The Journal of Rheumatology

Serologic Evidence of Gut-driven Systemic Inflammation in Juvenile Idiopathic

Arthritis
Lampros Fotis, Nurmohammad Shaikh, Kevin W. Baszis, Charles M. Samson, Raffi
Lev-Tzion, Anthony R. French and Phillip I. Tarr

DOI: 10.3899/jrheum.161589
http://www.jrheum.org/content/early/2017/09/11/jrheum.161589

1. Sign up for TOCs and other alerts

http://www.jrheum.org/alerts

2. Information on Subscriptions
http://jrheum.com/faq

3. Information on permissions/orders of reprints

http://jrheum.com/reprints_permissions

The Journal of Rheumatology is a monthly international serial edited by Earl D.

Silverman featuring research articles on clinical subjects from scientists working
in rheumatology and related fields.

Downloaded from www.jrheum.org on September 17, 2017 - Published by The

Journal of Rheumatology
Serologic Evidence of Gut-driven Systemic
Inflammation in Juvenile Idiopathic Arthritis
Lampros Fotis, Nurmohammad Shaikh, Kevin W. Baszis, Charles M. Samson, Raffi Lev-Tzion,
Anthony R. French, and Phillip I. Tarr
ABSTRACT. Objective. Accumulating evidence links juvenile idiopathic arthritis (JIA) to nonhost factors such as
gut microbes. We hypothesize that children with new-onset JIA have increased intestinal bacterial
translocation and circulating lipopolysaccharide (LPS).
Methods. We studied systemic treatment-naive patients with JIA [polyarticular JIA, n = 22, oligo-
articular JIA, n = 31, and spondyloarthropathies (SpA), n = 16], patients with established inflammatory
bowel disease–related arthritis (IBD-RA, n = 11), and 34 healthy controls. We determined circulating
IgG reactivity against LPS, LPS-binding protein (LBP), α-1-acid glycoprotein (α-1AGP), and
C-reactive protein (CRP) in plasma or serum from these patients and controls. Juvenile Arthritis
Disease Activity Score (JADAS-27) was calculated for patients with JIA.
Results. Circulating anticore LPS antibody concentrations in patients with polyarticular JIA
(p = 0.001), oligoarticular JIA (p = 0.024), and SpA (p = 0.001) were significantly greater than in
controls, but there were no significant intergroup differences. Circulating LBP concentrations were
also significantly greater in patients with polyarticular JIA (p = 0.001), oligoarticular JIA (p = 0.002),
and SpA (p = 0.006) than controls, as were α-1AGP concentrations (p = 0.001, 0.001, and 0.003,
respectively). No differences were observed between controls and patients with IBD-RA in any of the
assays. Circulating concentrations of LBP and α-1AGP correlated strongly with CRP concentrations
(r = 0.78 and r = 0.66, respectively). Anticore LPS antibody levels and CRP (r = 0.26), LBP (r = 0.24),
and α-AGP (r = 0.22) concentrations had weaker correlations. JADAS-27 scores correlated with LBP
(r = 0.66) and α-1AGP concentrations (r = 0.58).
Conclusion. Children with polyarticular JIA, oligoarticular JIA, and SpA have evidence of increased
exposure to gut bacterial products. These data reinforce the concept that the intestine is a source of
immune stimulation in JIA. (J Rheumatol First Release September 15 2017; doi:10.3899/jrheum.
161589)

Key Indexing Terms:

JUVENILE IDIOPATHIC ARTHRITIS ACUTE-PHASE PROTEINS
α1-ACID GLYCOPROTEIN INTESTINAL PERMEABILITY LIPOPOLYSACCHARIDE
LIPOPOLYSACCHARIDE-BINDING PROTEIN

Emerging data suggest that the clinical course of juvenile is manifest as lymphonodular hyperplasia and increased
idiopathic arthritis (JIA) might be influenced by nonhost tissue γ/δ+ and cytotoxic lymphocyte populations3,4.
factors, including gut microbes (reviewed1). These lines of Additionally, patients with JIA have defective intestinal
evidence associate microbial dysbiosis with altered gut barrier function5,6, and many children with JIA and
permeability and a proinflammatory extraintestinal cascade abdominal pain have evidence of microscopic colitis7. Also,
driven by gut processes that contribute to the development exclusive enteral nutrition, which induces remission in Crohn
of arthritis2. Intestinal immune activation in patients with JIA disease, has been beneficial in patients with JIA8, and associ-

From the Divisions of Rheumatology and Gastroenterology, Hepatology, St. Louis; C.M. Samson, MD, Assistant Professor of Pediatrics, Pediatric
and Nutrition, Department of Pediatrics, Washington University School of Gastroenterology, Washington University in St. Louis; R. Lev-Tzion, MD,
Medicine, St. Louis, Missouri, USA; Nottingham University Hospitals, UK Assistant Professor, Pediatric Gastroenterology, Shaare Zedek Medical
National Health Service (NHS) Trust, Nottingham, UK; Juliet Keidan Center; A.R. French, MD, PhD, Associate Professor of Pediatrics,
Institute of Pediatric Gastroenterology and Nutrition, Shaare Zedek Pediatric Rheumatology, Washington University in St. Louis; P.I. Tarr,
Medical Center, Jerusalem, Israel. MD, Professor of Pediatrics, Pediatric Gastroenterology, Washington
Supported by US National Institutes of Health Grants P30DK052574 University in St. Louis.
(DDRCC Biobank Core), UL1TR000448, and P30CA091842 (for Address correspondence to Dr. P.I. Tarr, Department of Pediatrics,
REDCap), and funding from the Melvin E. Carnahan Professorship (Dr. Washington University, Box 8208, 660 South Euclid Ave., St. Louis,
Tarr). Missouri 63110, USA, E-mail: [email protected]; or Dr. A.R. French,
L. Fotis, MD, PhD, Consultant Pediatric Rheumatologist, Nottingham Department of Pediatrics, Washington University, Box 8208,
University Hospitals; N. Shaikh, PhD, Staff Scientist, Department of 660 South Euclid Ave., St. Louis, Missouri 63110, USA.
Pediatrics, Washington University in St. Louis; K.W. Baszis, MD, Assistant E-mail: [email protected]
Professor of Pediatrics, Pediatric Rheumatology, Washington University in Accepted for publication June 21, 2017.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2017. All rights reserved.

Fotis, et al: Gut-driven inflammation in JIA 1

Downloaded from www.jrheum.org on September 17, 2017 - Published by The

Journal of Rheumatology
ations between gut microbiome alterations and JIA have been patients with JIA who were prospectively recruited. All plasmas were
reported9,10,11. However, these studies have generally been collected in EDTA tubes, centrifuged (4°C, 15 min, 2500 g), and frozen
(–80°C) in aliquots until tested.
performed on subjects with longstanding disease.
Lipopolysaccharide (LPS), an outer-membrane compon-
Escherichia coli core LPS preparation and characterization. E. coli strain
F12 (pSK+) is a TnphoA mutant of E. coli O157:H7 that cannot express the
ent of Gram-negative bacteria, consists of a highly conserved O157 LPS side chain26 and therefore can serve as a source of antigen with
lipid A and a core oligosaccharide, and diverse sero- which to measure reactivity to generic Enterobacteriaceae LPS. This mutant
group-specific sugar (O) side chains12. The colon harbors an was inoculated from frozen stock into 2 l of Luria-Bertani broth containing
abundance of LPS-producing bacteria, and absorption of LPS 100 μg/ml of ampicillin and placed in a shaking incubator (37°C, overnight).
from the gut may cause systemic inflammation13. LPS
Bacteria were pelleted, washed once gently with purified sterile water, and
suspended in 20 ml of sterile water per 3–4 g of pellet. The suspension was
activates macrophages and neutrophils, which then synthe- heated (68°C, 10 min) in a water bath with stirring. Then an equal volume
size proinflammatory cytokines that might initiate and/or of preheated (68°C) phenol was slowly added, and the resulting suspension
perpetuate joint inflammation and degeneration14. was stirred vigorously (68°C, 60 min), placed in an ice water bath with
LPS-binding protein (LBP) is upregulated by LPS and continuous stirring (10 min), and centrifuged (4°C, 45 min). The upper layer
augments innate immunity to bacterial infections. LBP-bound
was aspirated and saved on ice. The organic interface and phenol layer with
bacterial pellet were reextracted with an equal volume of purified water,
LPS is transferred by CD14 on monocytes to the Toll-like repeating the mixing, heating, cooling, and centrifugation steps as before.
receptor 4 (TLR-4)/myeloid differentiation factor 2 (MD-2) The upper layers from both extractions were pooled and dialyzed against
signaling complex, the activation of which triggers inflam- 4 l of purified water using Spectra/Por 7 dialysis membrane 1000 kD
mation15,16. Indeed, LBP has been implicated in joint inflam- MWCO (Spectrum Laboratories) for 2 days, with 2 water changes daily,
mation based on elevated intraarticular concentrations in
after which the dialysate was ultracentrifuged (105,000 g, 4 h). The super-
natant was removed and the LPS pellet was suspended in molecular biology
rheumatoid arthritis17. There is also α-1-acid glycoprotein grade de-ionized water (Corning). Following DNase (EZ Bioresearch) and
(α-1AGP), another acute-phase protein with immuno- proteinase K digestion, we mixed an equal volume of 1:1 phenol/chloroform
modulatory properties18 that neutralizes the toxicity of LPS solution and then chloroform to remove residual phenol27.
and enhances its clearance from the body19. The circulating LPS was separated electrophoretically in a 10% sodium dodecyl
concentration of α-1AGP is associated with disease activity
sulfate-polyacrylamide gel28, and visualized by silver staining (GelCode
SilverSNAP Stain kit), and quantified in endotoxin units (EU/ml) by the
in adult rheumatoid arthritis and inflammatory bowel disease chromogenic limulus amebocyte lysate assay (Pierce Biotechnology).
(IBD)20,21. E. coli anticore LPS antibodies enzyme immunoassay. The concentration of
These findings prompted us to seek evidence of increased circulating anticore LPS antibodies was determined by a labora-
intestinal permeability and/or reactivity to gut bacterial tory-developed enzyme immunoassay (EIA). Nunc Maxisorp microplates
contents in treatment-naive patients with JIA. Such (Thermo Scientific) were coated (4°C, overnight) with 1000 EU of LPS per
increased permeability and exposure to gut contents might well diluted in 100 μl of 0.1 M carbonate-bicarbonate buffer (pH 9.6;
Sigma). The plates were washed 4 times with phosphate-buffered saline
be reflected in antibodies directed against LPS and in (PBS; pH 7.4) containing 0.05% Tween-20 (PBS/T) followed by incubation
elevated concentrations of circulating LBP. We also postu- (room temperature, 2 h) with 100 μl of blocking solution [1% bovine serum
lated variability in host response between children with albumin (BSA) in PBS/T], and washed again 4 times with PBS/T. The wells
different JIA subtypes. were then incubated with 100 μl of subject serum or plasma (1:100 diluted
in 0.5% BSA in PBS/T; room temperature, 2 h). After washing 4 times with
PBS/T, anti-human IgG/Fcγ-horseradish peroxidase (Jackson Laboratory),
MATERIALS AND METHODS 1:10,000 in 0.5% BSA in PBS/T, was added (100 µl) to each well and
The study was approved by the institutional review board (IRB) at incubated (room temperature, 1 h). The plates were then washed as before
Washington University School of Medicine, St. Louis (IRB ID# 201408105). and incubated with 100 μl of o-phenylenediamine dihydrochloride (Sigma)
Patients with JIA were eligible for enrollment if they were systemic substrate solution (25 min) before reaction termination with 100 μl of
treatment-naive and had not received antibiotics for the prior month. Patients 2N H2SO4. Absorbance was measured at 490 nm with Molecular Devices
previously treated with intraarticular corticosteroids were included. JIA was VERSAmax tunable microplate reader (Conquer Scientific).
defined according to the International League Against Rheumatism criteria Each sample was measured in 3 independent experiments performed in
for classifying idiopathic arthritis of childhood22. Sera and plasma from our duplicate on different days. For each experiment, we included as a standard
pediatric rheumatology specimen bank were used as well. Patients with the serum of a patient with polyarticular JIA with strong reactivity against
polyarticular JIA, oligoarticular JIA, enthesitis-related arthritis (ERA), and core LPS. This reactivity was arbitrarily defined as 1000 EIA units.
psoriatic arthritis (PsA) were included, but those with systemic JIA were
LBP, α-1AGP, and CRP concentrations. Commercial EIA were used to
excluded, because this entity is now widely thought to be an autoinflam-
determine concentrations of circulating LBP (HyCult) and α-1AGP and CRP
matory condition23. Patients with ERA and PsA were considered together as
(R&D Systems), according to the manufacturers’ instructions.
spondyloarthropathies (SpA) for the purpose of this analysis. We also
included patients who had IBD-related arthritis (IBD-RA) of variable Statistics. The Shapiro-Wilk test was used to determine whether the data
duration. All patients with IBD-RA were receiving systemic treatment for were normally distributed. The nonparametric Levene’s test was used to
IBD at the time they were studied, and their joint symptoms were assessed verify the equality of variances in the samples. The Kruskal-Wallis test was
by a pediatric rheumatologist. Sera and plasma from healthy control subjects used for multiple between-group comparisons. Spearman rank-order corre-
were obtained from our pediatric rheumatology plasma bank, as well as from lations were used to determine whether age of control subjects was related
normal controls obtained during studies of the pathophysiology of childhood to resulting values. In the 10 pairwise comparisons between groups for the
hemolytic uremic syndrome24, and of children undergoing colonoscopy for variables in Table 1, we considered p values < 0.005 to be significant (i.e.,
suspected IBD, but whose colonoscopy did not demonstrate disease. The 0.05 ÷ 10) after correcting for multiple comparisons. After demonstrating
Juvenile Arthritis Disease Activity Score (JADAS-27)25 was calculated for statistically significant differences using uncorrected p values for single

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2017. All rights reserved.

2 The Journal of Rheumatology 2017; 44:11; doi:10.3899/jrheum.161589

Downloaded from www.jrheum.org on September 17, 2017 - Published by The

Journal of Rheumatology
Table 1. Patient demographics.

Variables Oligoarticular JIA, n = 31 Polyarticular JIA, n = 22 SpA, n = 16 IBD-related Arthritis, n = 11 Control Group, n = 34

Age, yrs, median (IQR) 4.0 (3.0–9.0)1 11.5 (9.8–14.0) 14.5 (10.5–17) 15.0 (10–16) 10.5 (6.8–15)
Sex, male:female, n (%) 11 (34):20 (65) 8 (36):14 (64) 10 (63):6 (37) 7 (64):4 (36) 19 (56):15 (44)
Race White = 30 White = 20 White = 13 White = 10 White = 32
African American = 1 African American = 1 African American = 3 African American = 1 Indian/Alaskan = 1
African American/ Unknown = 1
Asian/White = 1
Ethnicity Non-Hispanic = 31 Non-Hispanic = 22 Non-Hispanic = 16 Non-Hispanic = 11 Non-Hispanic = 33
Hispanic = 0 Hispanic = 0 Hispanic = 0 Hispanic = 0 Hispanic = 1
JADAS-27 score,
median (IQR) 8.85 (7.3–13.6)2 20 (13–26.9)3 12.75 (9–16.9)3 NA NA

1 The age of the oligoarticular JIA group was significantly lower than the ages of all other groups (uncorrected p value for all comparisons < 0.001, with signifi-

cance set at 0.005 after correction for multiple comparisons between all groups). None of the remaining comparisons were statistically significant. 2 Eight
subjects evaluated. 3 Eleven subjects evaluated. JADAS: Juvenile Arthritis Disease Activity Score; JIA: juvenile idiopathic arthritis; SpA: spondyloarthropathies;
IBD: inflammatory bowel disease; IQR: interquartile range; NA: not applicable.

comparisons between individual JIA subgroups (but not the IBD-RA group) provided in more detail in Supplementary Table 1 (available
and the controls for all values, we compared the assay results for the with the online version of this article).
polyarticular JIA to oligoarticular JIA, polyarticular JIA to SpA, and oligo-
articular JIA to SpA groups. For the 3 pairwise comparisons among the
We detected significant differences between the median
juvenile arthritis subgroups, we provide uncorrected p values, but consider EIA values for anticore LPS antibody concentrations
p values < 0.017 (i.e., 0.05 ÷ 3) as significant after correcting for multiple (Table 2 and Figure 1) of healthy controls and polyarticular
comparisons. Linear relationships between the variables were measured with JIA (p < 0.001), oligoarticular JIA (p = 0.02), and SpA
the Spearman’s ρ test. The family error was set to p < 0.05. SPSS 22.0 was (p = 0.001) groups. There was not a significant difference in
used for these analyses. All p values were 2-tailed.
circulating anticore LPS antibody concentrations between the
healthy controls and IBD-RA subjects. Circulating LBP
RESULTS concentrations (Table 2 and Figure 2) were significantly
Samples from 114 subjects were analyzed. After correcting greater in the polyarticular (p = 0.001), oligoarticular
for multiple comparisons, the only statistically significant (p = 0.002), and SpA (p = 0.006) groups than in healthy
differences in the demographic data were between the ages controls. The IBD-RA subjects’ circulating LPB concentra-
of the oligoarticular JIA group and all other groups (Table 1). tions were not significantly higher than those in the controls.
The clinical features of the patients with IBD-RA are Circulating α-1AGP concentrations (Table 2 and Figure 3)

Table 2. Concentrations (median, interquartile range) of circulating anti-LPS, LBP, α-1AGP, and CRP in different groups. All comparisons are between individual
patient groups and healthy controls for each assay. P values < 0.05 are considered significant.

Assays Oligoarticular JIA, n = 31 Polyarticular JIA, n = 22 SpA, n = 16 IBD-related Arthritis, n = 11 Healthy Controls, n = 34

Anticore LPS, EIA units, 333 471 415 318 239

n = 112*, IQR 227–472 263–676 303–771 216–398 143–325
p value vs controls 0.024 < 0.001 0.001 0.111 NA

LBP, ng/ml, n = 112* 9.4 18.7 13.5 11.8 5.3

IQR 4.3–22.9 9.4–34.7 6.7–14.7 3.6–17.9 1.8–9.0
p value vs controls 0.002 0.001 0.006 0.108 NA

α-1AGP, μg/ml, n = 114 1203 1398 1116 759 658

IQR 820–1468 1025–1843 760–13,867 630–991 523–908
p value vs controls 0.001 0.001 0.003 0.185 NA

CRP, μg/ml, n = 107* 2.4 12.3 2.1 0.9 0.07

IQR 0.6–4.2 1.1–64.2 0.3–4.5 0–4.1 0–0.7
p value vs controls 0.001 < 0.001 0.001 0.122 NA

* For some subjects, we had insufficient serum or plasma to perform all EIA, so n for some assays does not always equal 114. LPS: lipopolysaccharide; LBP:

LPS-binding protein; α-1AGP: α-1-acid glycoprotein; CRP: C-reactive protein; JIA: juvenile idiopathic arthritis; SpA: spondyloarthropathies; EIA: enzyme
immunoassay; IBD: inflammatory bowel disease; NA: not applicable.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2017. All rights reserved.

Fotis, et al: Gut-driven inflammation in JIA 3

Downloaded from www.jrheum.org on September 17, 2017 - Published by The

Journal of Rheumatology
 Figure 1. Circulating anticore LPS antibody concentrations across groups. * Outliers above Q3 + 1.5 IQR.
Open circles are outliers between Q3 and Q3 + 1.5 IQR. IQR: interquartile range; LPS: lipopolysaccharide;
EIA: enzyme immunoassay; IBD-RA: inflammatory bowel disease–related arthritis.

Figure 2. Circulating LBP concentrations across groups. Open circles are outliers between Q3 and Q3 + 1.5 IQR.
IQR: interquartile range; IBD-RA: inflammatory bowel disease–related arthritis.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2017. All rights reserved.

4 The Journal of Rheumatology 2017; 44:11; doi:10.3899/jrheum.161589

Downloaded from www.jrheum.org on September 17, 2017 - Published by The

Journal of Rheumatology
 Figure 3. Circulating α-1AGP concentrations across groups. Open circles are outliers between Q3 and
Q3 + 1.5 IQR. α-1AGP: α-1-acid glycoprotein; IQR: interquartile range; IBD-RA: inflammatory bowel
disease–related arthritis.

were also significantly greater in all 3 JIA arthritis subgroups JADAS-27 score calculated for the 29 patients with JIA
than in the healthy controls (p = 0.001 for the polyarticular prospectively enrolled correlated strongly with the LBP
and oligoarticular JIA groups; p = 0.003 for the SpA groups), (r = 0.66; p < 0.001) and α-1AGP (r = 0.58; p = 0.001)
but were not greater in the patients with IBD-RA than in the concentrations, but not with concentrations of circulating
healthy controls. CRP concentrations were significantly anticore LPS antibodies.
lower in the healthy controls than in all 3 JIA disease We were concerned that the age of the controls was at
subgroups (p = 0.001, p < 0.001, and p = 0.001 for oligoar- variance with the ages of the members of the oligoarticular
ticular JIA, polyarticular JIA, and SpA groups, respectively; JIA group. However, for all values determined, only the
Table 2). As with the values for anticore LPS antibodies, LBP, concentration of circulating antibodies to LPS changed
and α-1AGP, the CRP values did not differ between healthy significantly over age (rs = 0.5288, t = 3.52, df = 32;
controls and the IBD-RA group. p = 0.001). Therefore, for this variable, we compared concen-
Supplementary Table 2 (available with the online version trations of antibodies to LPS between 11 controls in the
of this article) presents pairwise comparisons among the 3 youngest control group tertile [median age 6.0 yrs,
JIA disease subgroups. Circulating LBP concentrations were interquartile range (IQR) 4.0–7.0] and the 31 subjects in the
lower in the oligoarticular group than in the polyarticular JIA oligoarticular JIA group (median age 4.0 yrs, 3.0–9.0). The
group (p = 0.02), and circulating α-1AGP concentrations median circulating antibody concentration was 124 (IQR
were lower in the oligoarticular JIA and the SpA groups 57.1–236.5) EIA units in the controls, compared to 333 (IQR
(p = 0.033 and p = 0.026, respectively) than in the 227–472) EIA units in the oligoarticular group (p < 0.001),
polyarticular JIA group. However, none of these differences confirming the significance of the higher value in the oligo-
retained statistical significance after correcting for multiple articular group after controlling for age effects.
comparisons.
The concentrations of circulating CRP and α-1AGP DISCUSSION
(r = 0.77; p < 0.001), CRP and LBP (r = 0.78; p < 0.001), and The gut is a major habitat of microbes and their by-products.
LBP and α-1AGP (r = 0.66; p < 0.001) were strongly corre- The seroreactivity of patients with JIA to anticore LPS, a
lated. The concentrations of anticore LPS antibodies and component of Gram-negative bacterial cell walls, suggests
α-1AGP (r = 0.22; p = 0.02), LBP (r = 0.24; p = 0.012), and that such organisms and their systemic absorption may explain
CRP (r = 0.26; p = 0.007) had weaker correlations. The at least part of the inappropriate immune activation in JIA.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2017. All rights reserved.

Fotis, et al: Gut-driven inflammation in JIA 5

Downloaded from www.jrheum.org on September 17, 2017 - Published by The

Journal of Rheumatology
Because the E. coli mutant used to produce the core LPS CRP correlated strongly with the other 2 acute-phase
antigen for the EIA lacks the LPS O-side chain, we conclude proteins, namely LBP and α-1AGP, but only weakly with the
that the reactivity reflects a response to generic Gram-negative anticore LPS antibodies. The fact that LBP is directly linked
LPS in which the core and lipid A moieties are conserved29, to LPS stimulation suggests that this acute-phase reaction is
and does not reflect exposure to E. coli O157:H7. initiated by products from the intestine. Alternatively,
We found comparatively greater concentrations of circu- systemic inflammation might increase intestinal permeability
lating LBP and α-1AGP in treatment-naive patients with and secondarily increase concentrations of circulating
polyarticular JIA, oligoarticular JIA, and SpA than in healthy anticore LPS antibodies and/or LBP40. Although we found a
controls, suggesting systemic inflammation. Although the strong correlation between LBP and CRP concentrations in
elevation of LBP (and of α-1AGP) could reflect acute-phase the circulation (not surprising because both proteins are
reaction, the simultaneous presence of elevated anticore LPS acute-phase reactants), we found only a weak correlation
antibodies and LBP and α-1AGP concentrations indicates between α-1AGP and CRP concentrations and circulating
that Gram-negative bacteria or their products have influenced anticore LPS antibodies. This suggests that anticore LPS
this upregulation for an extended interval before sampling, antibody production is independent of a systemic inflam-
because antibody responses are slower to develop than matory state, but present chiefly in the polyarticular JIA,
acute-phase reactants. Because the gut is the largest habitat oligoarticular JIA, and SpA groups. JADAS-27 correlated
of Gram-negative bacteria in the body, this finding implicates significantly both with the LBP and α-1AGP, but not with
the intestine in the induction of systemic inflammation. Our anticore LPS antibodies, which is not surprising because both
findings of elevated concentrations of circulating LBP and acute-phase markers are related to CRP.
anticore LPS antibodies in patients with SpA are also Even though differences between some of the medians in
consistent with the observation that many such patients have pairwise comparisons between JIA subgroups were statisti-
subclinical intestinal inflammation30. Also, there are recent cally significant using uncorrected p values, none of the assay
reports of histologically determined intestinal inflammation values differ significantly after correcting for multiple
in patients with JIA7. The same finding has been confirmed comparisons. Hence, our data do not suggest that these
in patients with SpA31, a group in which elevated levels of markers can differentiate between oligoarticular JIA,
acute-phase reactants and anti-LPS were also observed. polyarticular JIA, and SpA patients with confidence.
Circulating LPS plausibly initiates or perpetuates systemic We acknowledge several limitations to our study. First,
inflammation. LPS is the primary ligand of TLR-4, and LPS our databases did not permit us to identify reliably
upregulates TLR-4, which in turn activates nuclear factor-κB nonsteroidal antiinflammatory drug (NSAID) use prior to
and PI-3K pathways that exert proinflammatory activity14. enrollment, and it is possible that these agents altered
LPS in human joint cartilage induces matrix breakdown and intestinal permeability. It will be important to account for this
chondrocyte apoptosis, thereby degrading cartilage11. variable in future prospective studies that seek to corroborate
TLR-4–deficient mice are relatively resistant to LPS-induced our findings. Second, we used both sera and plasmas in the
arthritis and joint destruction32,33. In addition, patients with assays. However, plasma and serum CRP41 and antimicrobial
rheumatoid arthritis are sensitized to Prevotella copri34 or antibody concentrations42,43 correlate well, and LBP concen-
antigens common to Enterobacteriaceae, including LPS35. trations from sera and plasmas are used in multiple case series
Further, Yersinia enterocolitica O:3 LPS is detectable in the (review44).
synovial fluid for extended periods in patients with postin- The patients with IBD-RA warrant several comments.
fectious reactive arthritis36. However, the elevated concen- First, the inability to find a significant elevation of any of the
trations of anticore LPS and LBP in polyarticular JIA and tested circulating markers in the IBD-RA group compared to
oligoarticular JIA are novel findings, because little published healthy controls was unexpected, because LBP concentra-
evidence exists for increased intestinal permeability or tions in Crohn disease exceed those in ulcerative colitis and
dysfunction in these disorders5,6. Moreover, our data are in normal controls45, and CRP is often used to indicate IBD
derived from treatment-naive patients, thereby lending activity. However, the central tendencies for each of these
credence to a potential role for gut bacterial content in precip- values exceeded those in the controls, suggesting some
itating juvenile arthritis. degree of immune activation in the IBD-RA group. Second,
Our data suggest the potential utility of α-1AGP to follow patients with IBD are admonished to avoid NSAID, and
disease activity in patients with JIA. Levels of α-1AGP were several patients had high circulating concentrations of
elevated in all 3 groups, and interestingly, were correlated antibodies to E. coli LPS, suggesting that systemic reaction
with the concentrations of LBP, CRP, and JADAS-27. to gut LPS can occur without this inciting agent, and even
Although not well studied in inflammatory arthritis, α-1AGP without active gut disease. Third, the comparative nonreac-
might play a role in the mechanisms of joint erosions and also tivity of this cohort might be attributed to the universal use
be of value as a biomarker in predicting the potential risk of of immunosuppressant agents. It is not clear why arthritis
erosions37,38,39. developed under these circumstances, but we must consider

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2017. All rights reserved.

6 The Journal of Rheumatology 2017; 44:11; doi:10.3899/jrheum.161589

Downloaded from www.jrheum.org on September 17, 2017 - Published by The

Journal of Rheumatology
the alternative hypothesis that the pathogenesis of IBD-RA, 3. Kokkonen J, Arvonen M, Vähäsalo P, Karttunen TJ. Intestinal
unlike JIA, is not related to systemic inflammation. Indeed, immune activation in juvenile idiopathic arthritis and connective
tissue disease. Scand J Rheumatol 2007;36:386-9.
there is little overlap in serologic profiles between adults with 4. Arvonen M, Ikni L, Augustin M, Karttunen TJ, Vähäsalo P. Increase
IBD-RA and rheumatoid arthritis46. However, the small size of duodenal and ileal mucosal cytotoxic lymphocytes in juvenile
of the IBD-RA group and the heterogeneity of immunosup- idiopathic arthritis. Clin Exp Rheumatol 2010;28:128-34.
pressants used limit the interpretations that can be drawn 5. Weber P, Brune T, Ganser G, Zimmer KP. Gastrointestinal
from this group. symptoms and permeability in patients with juvenile idiopathic
arthritis. Clin Exp Rheumatol 2003;21:657-62.
Overall, our findings and emerging knowledge of altered 6. Picco P, Gattorno M, Marchese N, Vignola S, Sormani MP,
intestinal microbiota in patients with JIA suggest a potential Barabino A, et al. Increased gut permeability in juvenile chronic
role for the gut in JIA. Although the concept of intestinal arthritides. A multivariate analysis of the diagnostic parameters.
dysfunction cannot yet be generalized to all patients with JIA, Clin Exp Rheumatol 2000;18:773-8.
those that have evidence of intestinal dysfunction as indicated 7. Pichler J, Ong C, Shah N, Sebire N, Kiparrissi F, Borrelli O, et al.
Histopathological features of gastrointestinal mucosal biopsies in
by elevated anti-LPS or LBP levels could be candidates for children with juvenile idiopathic arthritis. Pediatr Res 2016;
further evaluation of their gut bacterial population. It would 79:895-901.
also be interesting to study the response of patients to 8. Berntson L, Hedlund-Treutiger I, Alving K. Anti-inflammatory
bacterial antigens in the context of JIA disease monitoring. effect of exclusive enteral nutrition in patients with juvenile
Notably, in our present study, patients with IBD-RA, all of idiopathic arthritis. Clin Exp Rheumatol 2016;34:941-5.
9. Tejesvi MV, Arvonen M, Kangas SM, Keskitalo PL, Pirttilä AM,
whom were treated with immune suppressants and most of Karttunen TJ, et al. Faecal microbiome in new-onset juvenile
whom had quiescent gut disease, have in aggregate low idiopathic arthritis. Eur J Clin Microbiol Infect Dis 2016;35:363-70.
anti-LPS and LBP levels. However, it is premature to use any 10. Aggarwal A, Sarangi AN, Gaur P, Shukla A, Aggarwal R. Gut
of these markers as a treatment target in JIA. Additionally, in microbiome in children with enthesitis-related arthritis in a
future studies, it will be worthwhile to obtain stool from developing country and the effect of probiotic administration. Clin
Exp Immunol 2017;187:480-9.
patients to study fecal markers of gut reactivity at instructive 11. Stoll ML, Kumar R, Morrow CD, Lefkowitz EJ, Cui X, Genin A, et
points in the JIA disease process, and to perform sequence al. Altered microbiota associated with abnormal humoral immune
analysis to determine the presence and extent of potential responses to commensal organisms in enthesitis-related arthritis.
dysbiotic gut communities, and correlate these values with Arthritis Res Ther 2014;16:486.
markers of systemic inflammation as used in our study. 12. Trent MS, Stead CM, Tran AX, Hankins JV. Diversity of endotoxin
and its impact on pathogenesis. J Endotoxin Res 2006;12:205-23.
The concurrent increase in the concentrations of anticore 13. de Punder K, Pruimboom L. Stress induces endotoxemia and
LPS antibody and LBP in the circulation of treatment-naive low-grade inflammation by increasing barrier permeability. Front
patients with polyarticular JIA, oligoarticular JIA, and SpA Immunol 2015;6:223.
suggest systemic exposure to gut contents. Circulating LPS 14. Lorenz W, Buhrmann C, Mobasheri A, Lueders C, Shakibaei M.
increases tight junction permeability in mice47, and a similar Bacterial lipopolysaccharides form procollagen-endotoxin
complexes that trigger cartilage inflammation and degeneration:
process is postulated in patients with IBD, where circulating implications for the development of rheumatoid arthritis. Arthritis
LPS concentrations correlate with tissue (gut) inflammation Res Ther 2013;15:R111.
and disease activity45,48,49. This exposure potentially induced 15. Gutsmann T, Müller M, Carroll SF, MacKenzie RC, Wiese A,
a systemic inflammatory reaction in our cohorts, as evidenced Seydel U. Dual role of lipopolysaccharide (LPS)-binding protein in
by increased circulating LBP and α-1AGP concentrations in neutralization of LPS and enhancement of LPS-induced activation
of mononuclear cells. Infect Immun 2001;69:6942-50.
all subgroups except IBD-RA (the only group in which 16. Berner R, Fürll B, Stelter F, Dröse J, Müller H-P, Schütt C. Elevated
patients were under treatment), with the majority being in levels of lipopolysaccharide-binding protein and soluble CD14 in
disease remission. The concentrations of the gut-related plasma in neonatal early-onset sepsis. Clin Diagn Lab Immunol
circulating acute-phase proteins (LBP and α-1AGP) correlate 2002;9:440-5.
well with disease activity scores of patients with JIA, and 17. Heumann D, Bas S, Gallay P, Le Roy D, Barras C, Mensi N, et al.
Lipopolysaccharide binding protein as a marker of inflammation in
therefore these molecules are worthy of further consideration synovial fluid of patients with arthritis: correlation with interleukin
in monitoring disease activity. 6 and C-reactive protein. J Rheumatol 1995;22:1224-9.
18. Hochepied T, Berger FG, Baumann H, Libert C. α1-Acid
ONLINE SUPPLEMENT glycoprotein: an acute phase protein with inflammatory and
Supplementary material accompanies the online version of this article. immunomodulating properties. Cytokine Growth Factor Rev
2003;14:25-34.
19. Moore DF, Rosenfeld MR, Gribbon PM, Winlove CP, Tsai CM.
REFERENCES Alpha-1-acid (AAG, orosomucoid) glycoprotein: interaction with
1. Arvonen M, Berntson L, Pokka T, Karttunen TJ, Vähäsalo P, Stoll bacterial lipopolysaccharide and protection from sepsis.
ML. Gut microbiota-host interactions and juvenile idiopathic Inflammation 1997;21:69-82.
arthritis. Pediatr Rheumatol Online J 2016;14:44. 20. Cylwik B, Chrostek L, Gindzienska-Sieskiewicz E, Sierakowski S,
2. Lerner A, Matthias T. Rheumatoid arthritis-celiac disease Szmitkowski M. Relationship between serum acute-phase proteins
relationship: joints get that gut feeling. Autoimmun Rev and high disease activity in patients with rheumatoid arthritis. Adv
2015;14:1038-47. Med Sci 2010;55:80-5.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2017. All rights reserved.

Fotis, et al: Gut-driven inflammation in JIA 7

Downloaded from www.jrheum.org on September 17, 2017 - Published by The

Journal of Rheumatology
21. Nielsen OH, Vainer B, Madsen SM, Seidelin JB, Heegaard NH. 37. Olewicz-Gawlik A, Korczowska-Łacka I, Łacki JK, Klama K,
Established and emerging biological activity markers of Hrycaj P. Fucosylation of serum alpha1-acid glycoprotein in
inflammatory bowel disease. Am J Gastroenterol 2000;95:359-67. rheumatoid arthritis patients treated with infliximab. Clin
22. Petty RE, Southwood TR, Baum J, Bhettay E, Glass DN, Manners Rheumatol 2007;26:1679-84.
P, et al. Revision of the proposed classification criteria for juvenile 38. Haston JL, FitzGerald O, Kane D, Smith KD. The influence of
idiopathic arthritis: Durban, 1997. J Rheumatol 1998;25:1991-4. alpha1-acid glycoprotein on collagenase-3 activity in early
23. Prakken B, Albani S, Martini A. Juvenile idiopathic arthritis. Lancet rheumatoid arthritis. Biomed Chromatogr 2003;17:361-4.
2011;377:2138-49. 39. De Graaf TW, Van Ommen EC, Van der Stelt ME, Kerstens PJ,
24. Chandler WL, Jelacic S, Boster DR, Ciol MA, Williams GD, Boerbooms AM, Van Dijk W. Effects of low dose methotrexate
Watkins SL, et al. Prothrombotic coagulation abnormalities therapy on the concentration and the glycosylation of alpha 1-acid
preceding the hemolytic-uremic syndrome. N Engl J Med glycoprotein in the serum of patients with rheumatoid arthritis: a
2002;346:23-32. longitudinal study. J Rheumatol 1994;21:2209-16.
25. Nordal EB, Zak M, Aalto K, Berntson L, Fasth A, Herlin T, et al. 40. Hietbrink F, Besselink MG, Renooij W, de Smet MB, Draisma A,
Validity and predictive ability of the juvenile arthritis disease van der Hoeven H, et al. Systemic inflammation increases intestinal
activity score based on CRP versus ESR in a Nordic permeability during experimental human endotoxemia. Shock
population-based setting. Ann Rheum Dis 2012;71:1122-7. 2009;32:374-8.
26. Bilge SS, Vary JC, Dowell SF, Tarr PI. Role of the Escherichia coli 41. Dossus L, Becker S, Achaintre D, Kaaks R, Rinaldi S. Validity of
O157:H7 O side chain in adherence and analysis of an rfb locus. multiplex-based assays for cytokine measurements in serum and
Infect Immun 1996;64:4795-801. plasma from “non-diseased” subjects: Comparison with ELISA.
27. Inzana TJ. Electrophoretic heterogeneity and interstrain variation of J Immunol Methods 2009;350:125-32.
the lipopolysaccharide of Haemophilus influenzae. J Infect Dis 42. Siev M, Yu X, Prados-Rosales R, Martiniuk FT, Casadevall A,
1983;148:492-9. Achkar JM. Correlation between serum and plasma antibody titers
28. Tsai CM, Frasch CE. A sensitive silver stain for detecting to mycobacterial antigens. Clin Vaccine Immunol 2011;18:173-5.
lipopolysaccharides in polyacrylamide gels. Anal Biochem 43. Cherpes TL, Meyn LA, Hillier SL. Plasma versus serum for
1982;119:115-9. detection of herpes simplex virus type 2-specific immunoglobulin G
29. Tyler JW, Cullor JS, Spier SJ, Smith BP. Immunity targeting antibodies with a glycoprotein G2-based enzyme immunoassay.
common core antigens of gram-negative bacteria. J Vet Intern Med J Clin Microbiol 2003;41:2758-9.
1990;4:17-25. 44. Gonzalez-Quintela A, Alonso M, Campos J, Vizcaino L, Loidi L,
30. Mielants H, Veys EM, Cuvelier C, De Vos M, Goemaere S, De Gude F. Determinants of serum concentrations of
Clercq L, et al. The evolution of spondyloarthropathies in relation to lipopolysaccharide-binding protein (LBP) in the adult population:
gut histology. II. Histological aspects. J Rheumatol 1995;22:2273-8. the role of obesity. PLoS One 2013;8:e54600.
31. Conti F, Borrelli O, Anania C, Marocchi E, Romeo EF, Paganelli M, 45. Pasternak BA, D’Mello S, Jurickova II, Han X, Willson T, Flick L,
et al. Chronic intestinal inflammation and seronegative spondy- et al. Lipopolysaccharide exposure is linked to activation of the
loarthropathy in children. Dig Liver Dis 2005;37:761-7. acute phase response and growth failure in pediatric Crohn’s disease
32. Lee EK, Kang SM, Paik DJ, Kim JM, Youn J. Essential roles of and murine colitis. Inflamm Bowel Dis 2010;16:856-69.
Toll-like receptor-4 signaling in arthritis induced by type II collagen 46. Van Erp SJ, Verheul MK, Levarht EW, van der Reijden JJ, van der
antibody and LPS. Int Immunol 2005;17:325-33. Heijde D, van Gaalen FA, et al. Short article: Absence of serological
33. Pierer M, Wagner U, Rossol M, Ibrahim S. Toll-like receptor 4 is rheumatoid arthritis biomarkers in inflammatory bowel disease
involved in inflammatory and joint destructive pathways in patients with arthropathies. Eur J Gastroenterol Hepatol
collagen-induced arthritis in DBA1J mice. PLoS One 2017;29:345-8.
2011;6:e23539. 47. Guo S, Nighot M, Al-Sadi R, Alhmoud T, Nighot P, Ma TY.
34. Pianta A, Arvikar S, Strle K, Drouin EE, Wang Q, Costello CE, et Lipopolysaccharide regulation of intestinal tight junction
al. Evidence for immune relevance of Prevotella copri, a gut permeability is mediated by TLR4 signal transduction pathway
microbe, in patients with rheumatoid arthritis. Arthritis Rheumatol activation of FAK and MyD88. J Immunol 2015;195:4999-5010.
2017;69:964-75. 48. Lakatos PL, Kiss LS, Palatka K, Altorjay I, Antal-Szalmas P, Palyu
35. Aoki S, Yoshikawa K, Yokoyama T, Nonogaki T, Iwasaki S, Mitsui E, et al. Serum lipopolysaccharide-binding protein and soluble
T, et al. Role of enteric bacteria in the pathogenesis of rheumatoid CD14 are markers of disease activity in patients with Crohn’s
arthritis: evidence for antibodies to enterobacterial common disease. Inflamm Bowel Dis 2011;17:767-77.
antigens in rheumatoid sera and synovial fluids. Ann Rheum Dis 49. Pastor Rojo O, López San Román A, Albéniz Arbizu E, de la Hera
1996;55:363-9. Martínez A, Ripoll Sevillano E, Albillos Martínez A. Serum
36. Granfors K, Merilahti-Palo R, Luukkainen R, Möttönen T, lipopolysaccharide-binding protein in endotoxemic patients with
Lahesmaa R, Probst P, et al. Persistence of Yersinia antigens in inflammatory bowel disease. Inflamm Bowel Dis 2007;13:269-77.
peripheral blood cells from patients with Yersinia enterocolitica O:3
infection with or without reactive arthritis. Arthritis Rheum
1998;41:855-62.

Personal non-commercial use only. The Journal of Rheumatology Copyright © 2017. All rights reserved.

8 The Journal of Rheumatology 2017; 44:11; doi:10.3899/jrheum.161589

Downloaded from www.jrheum.org on September 17, 2017 - Published by The

Journal of Rheumatology