Content uploaded by Mikael Hellström
Author content
All content in this area was uploaded by Mikael Hellström
Content may be subject to copyright.
1080
Interleukin (IL)-6 and IL-8 in Children with Febrile Urinary Tract Infection and
Asymptomatic Bacteriuria
Mikael Benson,
Vlf
Jodal, William Agace,
Mikael Hellstrom, Staffan Marild, Sten Rosberg,
Michael Sjostrom,
Bjorn
Wettergren, Susanne Jonsson,
and
Catharina
Svanborg
Divisions
of
Pediatric Nephrology and
of
International Pediatric
Growth Research Centre, Department
of
Pediatrics, East Hospital, and
Department
of
Radiology, Sahlgrenska Hospital, Goteborg University,
Goteborg; Department
of
Medical Microbiology, Division
of
Clinical
Immunology, Lund University, Lund; and Research Group for
Chemometrics, Department
of
Organic Chemistry,
Umed University, Umed, Sweden
Urine and serum interleukin (IL)-6and IL-8responseswerehigher in children with febrileurinary
tract infection
(n = 61) than in those with asymptomaticbacteriuria (n = 39).Byunivariate analysis,
cytokine levels were related to age, sex,
reflux,
renal scarring, urine leukocytes, C-reactive protein
(CRP), erythrocyte sedimentation rate (ESR), and bacterial properties (P fimbriae but not hemoly-
sin). Multivariate modeling showed that urine IL-6responseswerehigher in girlsthan boys,increased
with age, and were positively associated with CRP, ESR, serum IL-6, and urine leukocyte counts.
The urine IL-8 response was not influenced by age, but it was influenced by P fimbriae and was
associated with ESR, CRP, urine leukocytes, and female sex. The results show that cytokine re-
sponses to urinary tract infection vary with the severityof infection and that cytokine activation is
influenced by a variety of host and bacterial variables.
The symptoms and signs
of
urinary tract infection (UTI)
depend on the host response to the infecting bacterial strain
[1]. Local and systemic inflammatory changes cause fever and
the symptoms that characterize acute pyelonephritis. In patients
with asymptomatic bacteriuria (ABU), the bacteria do not gen-
erate a host response
of
sufficient magnitude to cause symp-
toms.
Escherichia coli strains that cause acute pyelonephritis
or
ABU
are known to differ in virulence [2], but the molecular
basis for the difference in host response
and
in clinical appear-
ance is not well understood.
Cytokines, such as interleukin (IL)-6 and IL-8, are activated
in patients with UTI
[3-12].
It
is speculated that cytokines are
mediators
of
the host responses to
UTI
and
that a difference
in magnitude or quality (or both)
of
the cytokine response
among patients with acute pyelonephritis and
ABU
underlies
Received 26 January 1996; revised 20 June 1996,
Presented in part: European Society for Pediatric Nephrology, Amsterdam,
Netherlands, September 1994 (abstract 78); International Business Communi-
cations 2nd Annual International Conference on Cytokine Therapy, Washing-
ton, DC, March 1995; International Pediatric Nephrology Association, Santi-
ago, Chile, August 1995 (abstract FC047); 3rd Annual Conference of the
International Cytokine Society, Harrogate, United Kingdom, September 1995
(abstract 362).
Informed consent was obtained from all parents, and the study was approved
by the Human Ethics Commmitte, Goteborg University, Goteborg, Sweden.
Grant support: Swedish Medical Research Council; Medical faculties of
Lund and G6teborg Universities;
Goteborg Medical Association; First of May
Flower Annual Campaign for Children's Health; Royal Physiographic Society;
Crawford and Osterlund Foundations.
Reprints or correspondence: Dr. Catharina Svanborg, Dept.
of
Medical Mi-
crobiology, Division
of
Clinical Immunology, Lund University, Solvegatan
23, S-223 62 Lund, Sweden.
The
Journal
oflnfectious
Diseases
1996;174:1080-4
© 1996 by The University of Chicago. All rights reserved.
0022-1899/96/7405-0025$01.00
the difference in clinical presentation [5]. The aims
of
the
present study were to compare the IL-6 and IL-8 responses
with UTI among children with acute pyelonephritis and ABU
and to analyze the influence
of
bacterial and host parameters
on the cytokine response using univariate and multivariate tech-
niques.
Patients
and
Methods
The study included 61 children from a prospective study of
febrile UTI [13] and 39 children with ABU from a screening study
[14]. To be included in the febrile group, children (2 months to 6
years old) had to be experiencing their first known symptomatic
UTI episode, have a fever of at least 38.5°Cwithin 24 h of diagno-
sis, and have bacteriuria, as determined by cultureof urine obtained
by suprapubic bladder aspiration (any growth), by uniform growth
of at least 10
5
bacteria/mL in 2 consecutive urine samples, or by
growth of at least 10
5
bacteria/mL in 1 urine sample and a positive
nitrite test. Urine samples for cytokine analysis were available
from 48 febrile children, and serum samples were available from
38. ABU was detected by screening 3581 children at 2 weeks, 3
months, and 10 months of age and was confirmed by culture of
urine obtained by suprapubic bladder aspiration. Samples for cyto-
kine analysis were available from 39 children with ABU (9 girls
and 30 boys).
Reflux was detected by cystourethrography [15] in 18 children
with acute pyelonephritis (4 with grade I, 10 with grade 2, and 4
with grade 3 reflux on a 5-grade scale) and in 4 others with ABU
(2 with grade I and 2 with grade 2). Renal scarring was detected
by urography [16] in 2 children with acute pyelonephritis at the
time of diagnosis and in another 7 children at follow-up (new
scarring). No scarring was detected in the ABU group at inclusion
or follow-up.
E. coli were isolated from children in 59 febrile episodes; the
remaining 2 children were infected with
Klebsiella and Enterococ-
by guest on July 13, 2011jid.oxfordjournals.orgDownloaded from
JIO 1996;174 (November)
Cytokines in Symptomatic and Asymptomatic UTI
1081
Table 1. Cytokine responses and host and bacterial variables in children with UTI.
Asymptomatic
bacteriuria
Febrile UTI
P
Urine IL-6 (U/mL), n
39
48
0(0)
28
(0-760)
<.001
Serum IL-6 (U/mL), n
30
38
o
(0-40)
10
(0-500)
<.001
Urine IL-8 (pg/mL), n
37
18
o
(0-380)
88
(0-3870)
<.01
Serum IL-8 (pg/mL), n
26
ND
0(0-128)
Age (years)
0.2 (0.1-0.9)
0.9
(0.2-5.8)
<.001
Girls/boys 9/30
44/17
<.001
Reflux
(+/-)
4/33 18/43
<.05
New scarring
(+/-)
0128
7/51
Urine leukocytes (cells/
J.lL)
22 (0-4000)
700 (11-10,000)
<.001
CRP (mg/L)
1
(1-21)
100
(1-200)
<.001
ESR (mm/h) 9
(1-40)
45
(2-72)
<.001
No. of Escherichia coli
32 59
P fimbriae
(+/-)
9/23 46/13 <.001
Hemolysin
(+/-)
13/19
36/23
NOTE. Except as indicated by
+/-
(present/absent) and for girls/boys, data are median (range). CRP, C-reactive
protein; ESR, erythrocyte sedimentation rate.
cus species, respectively. E. coli caused 32
of
the episodes in
children with ABU; 2 were infected with
Proteus species, 1 with
Staphylococcus aureus, 3 with Klebsiella species, and 1 with En-
terococcus
species. P fimbriae were identified by the ability
of
the
bacteria to induce P blood group-dependent mannose-resistant
agglutination of human erythrocytes [17, 18]. Hemolysin produc-
tion was assessed in nutrient agar with 5% washed horse erythro-
cytes.
The host response to UTI was analyzed in blood and urine
samples obtained at study entry. Erythrocyte sedimentation rate
(ESR, millimeters per hour) and C-reactive protein (CRP, milli-
grams per liter) were quantitated. Leukocytes in uncentrifuged
urine were counted using a Fuchs-Rosenthal chamber. The IL-6
activity in serum and urine samples was determined using the B9
bioassay with neutralizing
anti-IL-6
antibodies [19]. Urine IL-6
levels were confirmed by ELISA, based on the M16 monoclonal
anti-human IL-6 antibody, with polyclonal antibodies for detection
[20]. Serum IL-6 levels were confirmed using the Medgenix test
(Medgenix Diagnostics, Fleurus, Belgium). IL-8 levels in serum
and urine were determined by ELISA (reagents were provided by
M. Ceska, Sandoz, Vienna) [21]. IL-6 levels
~20
U/mL and IL-
8 levels ?:30 pg/mL were considered positive (responders) [4, 6].
The Savage score method [22] and Fisher's exact test were used
for univariate comparisons. Spearman's rank correlation test was
used to compute univariate correlations. Principal component anal-
ysis [23] and partial least squares to latent structures [24] were
used for multivariate analysis.
Results
Children with febrile
UTI
had higher IL-6 levels in urine
and serum and higher urine IL-8 levels than the children with
ABU (table 1). A urine IL-6 response (:?:20 U/mL) occurred
in 30 (63%)
of
48 children with febrile
UTI
and
in a
of
39
with
ABU
(P < .01). Elevated urine IL-8 levels
(~30
pg/
mL)
were
found in 13 (76%)
of
17 children with febrile
UTI
compared with 11 (30%)
of
37 children
with
ABU
(P < .01).
There was a serum IL-6 response in 19
of
38 children
with
febrile UTI,
but
only
1 (3%)
of
30 with
ABU
(P < .01) had
a response. Five
(19%)
of
26 children with
ABU
had
a serum
IL-8 response.
Serum
for analysis
of
IL-8 was not available
from subjects in
the
febrile group.
The patients
with
ABU
differed from those with febrile
UTI
by age, sex, reflux, inflammatory response,
and
properties
of
the infecting E. coli strain (table 1). By univariate analysis,
these variables all
had
a significant association
with
the cyto-
kine response. In the
combined
febrile and
ABU
groups, urine
IL-6 levels increased
with
age (P < .001). The
median
urine
IL-6 levels
were
significantly
higher
in girls
than
in boys (P
< .001) and in children with reflux (P < .001) or
new
renal
scarring
(P < .05)
than
in those with radiologically normal
urinary tracts.
Of
49 children infected with P fimbriated E.
coli, 23 (47%)
had
elevated urine IL-6 levels compared
with
6 (20%)
of
30 children infected
with
other E. coli strains (P
< .01). There
was
a correlation
of
urine IL-6 levels to urine
leukocytes, CRP,
and
ESRs
(all P < .001).
In the febrile group, the urine IL-6 levels increased with age
(P < .01) and
were
higher in girls (P < .01) and in children
with reflux
(P < .01) than in boys or children without reflux,
respectively. In the
ABU
group, serum IL-6 was correlated to
CRP
(P < .01).
Univariate analysis
of
the combined study group showed that
serum IL-6 levels increased with age
(P < .01). The median
by guest on July 13, 2011jid.oxfordjournals.orgDownloaded from
1082
Bensonet al.
.TID
1996;174 (November)
1[2]
p[2]
~-------------t[1]
p[1]
RP
0.40.2
0.0-0.2
B
.UIL
8
• PFIM
.HEMO
• SR
• ULEU
.MALE
.c
•
SIL6
• UIL6
•
REFLU~
AGE
• SCAR
0.0
0.4
0.2
-0.4
-0.2
t[3]_
~A------------------
I
I
I
Figure
1. A, 3-dimensional projection
of
febrile UTI (filled symbols) and asymptomatic bacteriuria (ABU); (open symbols) based on 12 study
variables: age, sex, reflux, renal scarring, C-reactive protein (CRP), erythrocyte sedimentation rate (SR), leukocytes, P fimbriae, hemolysin, serum
IL-6, urine IL-6, and urine IL-8. Fever was diagnostic criterion and therefore excluded from analysis. B, 2-dimensional, variable-oriented projection
of
combined febrile UTI and ABU groups. UIL8 = urine IL-8, PFIM = P fimbriae, HEMO = hemolysin, ULEU = urine leukocytes, SIL6 =
serum IL-6, UIL6 = urine IL-6, SCAR = renal scar.
t[l]
to t[3] are first 3 principal components, that is, t[I] is combination of variables that
explains largest part of variance, t[2], orthagonal to
t[l],
is another combination of variables that explains largest part of residual variance, and
so on. Loading scores
p[l]
and p[2] are coefficients with which variables are combined to form principal components t[l] and [2].
serum IL-6 levels were higher in girls than in boys (P < .01)
and higher
in children with reflux (P < .01) or in those with new
scarring
(P < .01) than in those without radiologic abnormalities.
There was a correlation between serum IL-6 levels and urine
leukocytes
(P < .01), ESRs (P < .05), and CRP (P < .001). In
the febrile group, serum IL-6 levels were elevated in children
with new scarring
(P < .05) and reflux (P < .01) but showed
no association with the other study variables.
Univariate analysis
of
the combined study groups showed
that the median urine IL-8 levels were higher in girls than in
boys
(P < .05) and higher in children infected with P fimbriae-
positive
E. coli (P < .01) than with other E. coli strains. The
urine IL-8 levels showed a correlation to urine leukocytes
(P
< .01), CRP (P < .01), and ESRs (P < .01). Urine IL-8
showed a poor association with the study variables when the
febrile and ABU groups were analyzed separately.
The data matrix with 12 variables (age, sex, reflux, new
scarring, urine leukocytes, urine IL-6, urine IL-8, serum IL-6,
CRP, ESR, P fimbriae, hemolysin) was reassessed using princi-
pal component analysis (figure 1A). The children with ABU
formed a tight cluster, well separated from the febrile group.
In contrast, two subsets were distinguished in the febrile group;
a two-dimensional, variable-orientedprincipal componentanal-
ysis was used to resolve the variables characteristic
of
the two
clusters (figure 1B). One cluster was characterized by infection
with P fimbriated and hemolysin-positive
E. coli, by elevated
levels of urine IL-8 and leukocytes, and elevated ESRs. The
other cluster was characterized by the absence
of
these bacterial
parameters, elevated urine and serum IL-6 levels, reflux, scar-
ring, and older age. CRP did not further contribute to the
separation of these groups.
Partial least
squares-
to-latent
structures analysis on the
combined data set (ABU and febrile UTI) showed a significant
association between urine IL-6 and CRP, female sex, ESR,
serum IL-6, age, and urine leukocytes. A model with these
variables explained 47%
of
the variation in urine IL-6. Urine
IL-6 in the febrile group was influenced by female sex, reflux,
serum IL-6, and age, explaining 32%
of
the variation. In the
combined data set, urine IL-8 showed an association with ESR,
P fimbriae, CRP, urine leukocytes, and female sex, but these
variables explained only 13%
of
the variation in urine IL-8.
Reflux was negatively correlated with urine IL-8. No significant
models for urine IL-8 were found when the ABU and febrile
groups were analyzed separately. Analysis of variables influ-
encing serum IL-6 and serum IL-8 levels gave no significant
models in any
of
the groups.
Discussion
Infections of the urinary tract activate local and systemic
cytokine responses
[3-12].
This study demonstrated that the
by guest on July 13, 2011jid.oxfordjournals.orgDownloaded from
JID 1996; 174 (November)
Cytokines in Symptomatic and Asymptomatic UTI
1083
serum IL-6, urine IL-6, and urine IL-8 levels were higher in
children with febrile UTI than in those with ABU. The clear
correlation between cytokine levels and disease severity sug-
gested that the cytokine response provides a partial explanation
for the link between bacterial infection, inflammation, and the
severity
of
UTI.
Children with ABU and febrile UTI differ in age, sex, reflux,
and renal scarring [25]. In addition, there are differences in
virulence
of
the causative E. coli strains and in the responses
to infection (measured as fever, CRP, ESR, and leukocytes in
blood and urine). In the present study, we observed that the
cytokine responses followed the background variables associ-
ated with acute pyelonephritis. IL-6 increased with age and
was higher in girls and children with reflux or renal scarring
than in boys or children with radiologically normal urinary
tracts. These results show that the same host variables are
associated with increased cytokine responses and acute pyelo-
nephritis. We speculate that persons with a tendency to respond
with high cytokine levels are more likely to develop acute
pyelonephritis. Conversely, the ABU group may include per-
sons who are low cytokine responders and who consequently
remain asymptomatic despite the presence
of
bacteria in the
urinary tract.
Both IL-6 and IL-8 are important early mediators of in-
flammation. IL-6 is an endogenous pyrogen, activator of acute-
phase reactants, including CRP, and a maturation factor for
mucosal lymphocytes. IL-8 is a chemoattractant for neutro-
phils. Release of cytokines from the site of infection precedes
the onset of fever, acute-phase responses, and neutrophil re-
sponses [6]. Despite the difference in kinetics
of
the responses,
we had expected to find higher IL-6 and IL-8 levels in patients
with fever, acute-phase reactants, or marked urine neutrophil
responses. Univariate and multivariate analysis
of
the combined
febrile and ABU groups showed that urine IL-6, serum IL-6,
and urine IL-8 levels were significantly related to CRP, ESRs,
and leukocyte counts. These associations did not remain sig-
nificant when the children with febrile UTI or ABU were ana-
lyzed separately, suggesting that the differences were valid for
the groups but not for individual patients.
We observed several differences between the IL-6 and IL-
8 host response patterns. IL-6 but not IL-8 varied with age,
renal scarring, and reflux. Principal component analysis sug-
gested that the febrile UTI group could be separated into two
subsets according to the two-dimensional, variable-oriented
projection. One subset consisted
of
younger P fimbriated E.
coli-infected children with high IL-8 and leukocyte levels and
high ESRs. This subset probably contains children who have
uncomplicated febrile UTI, experience single episodes of infec-
tion caused by bacteria
of
high virulence, are treated, and rarely
go on to develop renal scarring. The neutrophil-dominated in-
flammatory response in such patients may even contribute to
the elimination of bacteria from the urinary tract [6]. The sec-
ond subset of children was older, had elevated IL-6 levels,
vesicoureteric reflux, renal scarring, and were infected with P
fimbriae-negative and nonhemolysin-producing bacteria
more often than the first group. This subset may contain the
patients with vesicoureteric reflux who develop recurrent acute
pyelonephritis and run a higher risk to develop renal scarring.
This study shows that cytokine responses to UTI vary
with the severity
of
infection and that cytokine activation is
influenced by a variety
of
host and bacterial variables.
Taken
together, these results indicate that cytokine measurement
can be used to further discriminate between patients with
symptomatic UTI and ABU and to define patient groups for
further study.
Acknowledgment
We thank Bjorn Areschough at the Goteborg University Com-
puter Center for statistical advice.
References
1. Agace W, Connell H, Svanborg C. Host resistance to urinary tract infec-
tion. In: Urinary tract infections: molecular pathogenesis and clinical
management. Washington, DC: Mobley and Warren,
1995:221-43.
2. Johnson JR. Virulence factors in E. coli urinary tract infection. Clin Micro-
bioi Rev
1991;4:80-128.
3. de Man P, van Kooten C, Aarden L, Engberg I, Svanborg-Eden C. Interleu-
kin-6 induced at mucosal surfaces by gram-negative bacterial infection.
Infect Immun
1989;57:3383-8.
4. Hedges S, Andersson P, Lidin-Janson G, de Man P, Svanborg C. Interleu-
kin-6 response to deliberate colonization
of
the human urinary tract
with gram-negative bacteria. Infect Immun 1991;
59:421-7.
5. Hedges S, Stenqvist K, Lidin-Janson G, Martinell J, Sandberg T, Svanborg
C. Comparison of urine and serum concentrations
of
interleukin-6 in
women with acute pyelonephritis and asymptomatic bacteriuria. J Infect
Dis 1992;
166:653-6.
6. Agace W, Hedges SR, Ceska M, Svanborg C. Interleukin-8 and the neutro-
phil response to mucosal gram-negative infection. J Clin Invest 1993;
92:780-5.
7. Ko YC, Mukaida N, Ishiyama S, et al. Elevated interleukin-8 levels in the
urine of patients with urinary tract infections. Infect Immun 1993; 6I:
1307-
14.
8. Nicolle LE, Brunka J, Orr P, Wilkins J, Harding GK. Urinary immunoreac-
tive interleukin-l alpha and interleukin-6 in bacteriuric institutionalized
elderly subjects. J Urol 1993; 149:1049-53.
9. Tullus K, Fituri
0,
Burman LG, Wretlind B, Brauner A. Interleukin-6 and
interleukin-8 in the urine of children with acute pyelonephritis. Pediatr
Nephrol 1994;
8:280-4.
10. Petersson C, Hedges S, Stenqvist K, Sandberg T, Connell H, Svanborg C.
Suppressed antibody and interleukin-6 responses to acute pyelonephritis
in pregnancy. Kidney Int
1994;45:571-7.
II.
Jacobson SH, Hylander B, Wretlind B, Brauner A. Interleukin-6 and in-
terleukin-S in serum and urine in patients with acute pyelonephritis in
relation to bacterial virulence-associated traits and renal function. Neph-
ron
1994;67:172-9.
12. Benson M, Jodal U, Andreasson A, Karlsson A, Rydberg J, Svanborg C.
Interleukin 6 response to urinary tract infection in childhood. Ped Infect
Dis J 1994;
13:612-6.
13. Marild S, Wettergren B, Hellstrom M, Jodal U, Lincoln K, Orskov F,
Svanborg-Eden C. Bacterial virulence and inflammatory response in
infants with febrile urinary tract infection or screening bacteriuria. J
Pediatr 1988; 112:348-54.
by guest on July 13, 2011jid.oxfordjournals.orgDownloaded from
1084
Benson et al.
JID 1996; 174 (November)
14. Wettergren B, Jodal U, Jonasson G. Epidemiology
of
bacteriuria during
the first year
of
life. Acta Pediatr Scand 1985;
74:925-33.
15. Lebowitz RL, Olbing H, Parkkulainen KU, Smellie JM, Tamrninen-Mob-
ius TE. International system
of
radiographic grading
of
vesicoureteric
reflux. International Reflux Study in Children. Pediatr Radiol 1985; 15:
105-9.
16. Claesson I, Jacobsson B, Olsson T, Ringertz H. Assessment of renal paren-
chymal thickness in normal children. Acta Radiol [Diagn] 1981;22:
305-14.
17. Leffler H, Svanborg-Eden C. Chemical identification
of
a glycosphingo-
lipid receptor for
Escherichia coli attaching to human urinary tract
epithelial cells and agglutinating human erythrocytes. FEMS Microbiol
Lett
1980;8:127-34.
18. Leffler H, Svanborg-Eden C. Glycolipid receptors for uropathogenic Esch-
erichia coli
on human erythrocytes and uroepithelial cells. Infect Immun
1981;
34:920-9.
19. Helle M,
Boeije
L,
Aarden
LA.
Functional
discrimination be-
tween
interleukin-6
and
interleukin-l.
Eur J Immunol 1988; 18:
1535-40.
20. Hang L, Svanborg C, Andersson G. Determination of interleukin-6 in
human urine and epithelial cell supernatants. Int Arch Allergy Immunol
1994; 105:397
-403.
21. Martich GD, Danner RL, Ceska M, Suffrendini AF. Detection of interleukin-
8 and tumor necrosis factor in normal humans after intravenous endotoxin:
the effect of antiinflammatory agents. J Exp Moo 1991;173:771-4.
22. Hajek
1. A course in non-parametric statistics. San Francisco: Holden-Day,
1969.
23. Wold S. Principal component analysis. Chemometrics and intelligent labora-
tory systems, 2. Amsterdam: Elsevier Publications, 1987.
24. Hoskuldsson
A. PLS regression methods. J Chemometrics 1988;2:211-28.
25. Kunin
C. Detection, prevention and management of urinary tract infections.
Philadelphia: Lea and Febiger, 1987.
by guest on July 13, 2011jid.oxfordjournals.orgDownloaded from