Background

Inflammatory bowel disease (IBD), Crohn’s disease (CD) and ulcerative colitis (UC) are chronic intestinal disorders characterised by an often early onset in children and adolescents. The current prevalence of IBD is estimated to be 2–3/1,000 in Western and developed countries.

IBD is regarded as part of a broad spectrum of (auto)immune disorders characterised by failure to limit the inflammatory response to luminal antigens. In contrast with true autoimmune diseases, where auto-antibodies react with self antigens [5], inflammation in CD is directed against antigens that are mainly derived from the environment. The working hypothesis is that chronic IBD occurs as a result of a breakdown of tolerance to luminal bacteria in the digestive tract in a genetically predisposed host [1].

It is a general observation that the relative contribution of genetic (inherited) and environmental (acquired) factors are shifted more towards genes in younger patients and towards the environment in elderly patients. Therefore, the participation of young children in epidemiologic and genetic research is crucial in order to understand the role of genetics and environmental factors in the development of IBD. Since genetic variations over a 50-year span are negligible, the remarkable increase in the incidence of CD suggests that environmental factors are involved. A unifying hypothesis to explain the simultaneous increase in autoimmunity and IBD (Th1-mediated) and allergies (Th2-mediated) is that modern living conditions can lead to the defective maturation of regulatory T cells (Treg) and antigen-presenting cells (APCs).

The NOD2/CARD15 gene, an intracellular receptor for bacterial muramyldipeptide, has been associated with an increased susceptibility to CD, as well as to atopic eczema and asthma [23, 39]. Recently, an impaired dendritic cell function in patients with the NOD2 frameshift 1007insC mutation has been reported [24].

Genetic studies in IBD have highlighted associations with pattern recognition receptors (PRRs), of which CARD15 and toll-like receptor (TLR) 4 have been the most widely replicated [3, 4, 9, 10, 15, 17, 21, 30, 31, 36]. These studies further suggest that these genetic variants influence the disease location, behaviour and natural history. However, only a few paediatric studies have investigated markers of disease progression and severe disease course. CARD15 is a member of a family of human cytosolic PRRs, which are involved in inborn or innate immunity. PRRs recognise pathogen-associated molecular patterns of micro organisms through leucine-rich repeat (LRR)-containing domains and are essential to maintain an equilibrium between the host and the enteric flora and to counteract early bacterial invasion. CARD15 is a PRR that, contrarily to the cell-surface expressed TLRs, comes to expression in the cell cytoplasm and recognises muramyl dipeptide derived from Gram-positive bacteria [7, 16]. CARD15 is known to be a genetic risk factor for CD only and analysis of the genotype/phenotype relationship has shown that ileal inflammation is the rule in CARD15-mutated CD patients [2, 12, 18–20, 25, 26, 37].

The hypothesis that bacterial agents play an important role in the pathogenesis of IBD in a genetically predisposed individual is supported by the presence of antibody responses against bacterial sequences, of which antibodies against the oligomannan of Saccharomyces cerevisiae (ASCA) and against a yet undefined antigen present in neutrophils (ANCA), are the most common [33]. Louis et al. [27, 28] previously demonstrated that 46% of CD patients will undergo changes in the disease phenotype behaviour with increasing disease duration, with the major determinant of change being disease location. Ileal disease distribution has been associated with more complicated (stenotic and/or penetrating) disease. An association of ileal disease and ASCA has also been documented, but it is unclear whether the primary association of disease progression is with disease location or ASCA status.

In this study, we wanted to investigate the progression of paediatric IBD and to see if factors associated with more rapid progression could be identified. Second, given the observation that NOD2/CARD15 variants are found with increased prevalence in immune-mediated diseases other than CD [32], we investigated if concomitant immune-mediated diseases could be identified in IBD patients and if these individuals more frequently carry NOD2 or other innate immune receptor variants. We therefore collected all IBD patients with an onset of disease at <17 years of age who were in follow up at our hospital, as well as their first-degree relatives, and we studied them for clinical, genetic (CARD, TLR4) and serologic (ASCA, pANCA) markers.

Patients and methods

Patients and phenotypic analysis

Between 2004 and 2005, 99 consecutive patients with a diagnosis of IBD before the age of 17 years and followed at our hospital (a tertiary referral centre) were included. The majority of patients were diagnosed with CD (n = 80), 15 with UC and four with indeterminate colitis. All individuals were of white Caucasian origin, except for two patients, who were Moroccan. The diagnosis of CD or UC was based on standard clinical, endoscopic, radiological and histological criteria [8, 22].

We also included 335 healthy first-degree family members of these children (190 parents and 145 siblings) and a control group of 76 children without IBD or other inflammatory conditions who consulted the paediatric gastroenterology unit of our hospital. All patients and/or parents (in children aged below 18 years) gave written informed consent to take part in this study.

The following clinical characteristics at the time of diagnosis were retrieved through patient chart review: age at diagnosis, gender, family history, order of birth, disease location, disease behaviour, need for abdominal surgery and the presence of concomitant immune-mediated diseases (Table 1). Disease extent was systematically assessed in all children by the use of upper and lower gastrointestinal endoscopy and small-bowel follow through. The following clinical characteristics were also analysed at maximal follow up: disease location, disease behaviour, need for abdominal surgery and the presence of concomitant immune-mediated diseases.

Table 1 Patient characteristics
Full size table

Before commencement of the study, the Ethics Committee of Gasthuisberg, Leuven, Belgium, approved the recruitment protocols.

Genotyping

All blood samples were taken at the time of consent and enrolment. DNA and serum were extracted from whole venous blood and stored at –80°C. All individuals were analysed for CARD15 R702W, G908R and L1007fs, for TLR4 Asp299Gly and for ASCA and pANCA.

CARD15 genotyping

All individuals were genotyped for the three CD-associated single-nucleotide polymorphisms (SNPs) in the CARD15 gene: Arg702Trp (GenBank accession number G67950), Gly908Arg (GenBank accession number G67954) and Leu1007fsinsC (GenBank accession number G67955). Genotyping analyses were performed using restriction fragment length polymorphism (RFLP) techniques. For Arg702Trp (amplification primers forward AGATCACAGCAGCCTTCCTG and reverse CACGCTCTTGGCCTCACC), presence of the mutation leads to the loss of one of the three restriction sites for MspI. Gly908Arg was amplified using forward primer CCCAGCTCCTCCCTCTTC and reverse primer AAGTCTGTAATGTAAAGCCAC, followed by an HhaI digestion, resulting in an intact fragment for the wild type (380 bp) or two fragments (138 bp and 242 bp) for the variant. The insertion mutation Leu1007insC (forward GGCAGAAGCCCTCCTGCAGGGCC and reverse CCTCAAAATTCTGCCATTCC) creates a restriction site for ApaI in the 151-bp fragment.

TLR4 genotyping

The principle of the method is based on the analysis of melting curves generated by Taqman probe hybridisation on a polymerase chain reaction (PCR) preamplified region of interest, in an “all-in-one” reaction. Primers and probes were designed with the freeware Melcalc v2.0 demo (http://www.melcalc.de) and purchased by Eurogentec (Brussels, Belgium). Briefly, for detection of the D299GTLR4 polymorphism, 100 nM of 5’ CTA GAG GGC CTG TGC AAT TTG ACC-3’ (forward primer), Fam-ACT ACT ACC TCG ATG GTA-DarqQuencher (probe) and 900 nM of ‘5 CCC TTT CAA TAG TCA CAC TCA CCA GG-3’ (reverse primer) were combined with 1xPlatinum Quantitative SuperMix-UDG (Invitrogen, Merelbeke, Belgium) and 100 ng of patient DNA in a 25-ul reaction volume. For detection of the CD14 ‘’C-260 T” promoter polymorphism, 100 nM of 5’ CCC TTC CTT TCC TGG AAA TAT TGC A-3’ (reverse primer), Fam-GGA CCG TAA CAG GAA G-DarqQuencher (probe) and 900 nM of ‘5 TGC CAG GAG ACA CAG AAC CCT AGA T 3’ (forward primer) were used under the same conditions as described above. The PCR conditions were as follows: step 1, 50°C (2 min); step 2, 95°C (2 min); step 3 (40×), 95°C (10 s), 65° (30 s) and 72°C (20 s); step 4, 95°C (2 min); step 5, 20–70°C (ramping time 20 min). All experiments were performed on 7700 Taqman apparatus (Applied Biosystems, Anderlecht, Belgium). Information collected during the melting curve step was analysed with SDS software version 1.7. The multicomponent file obtained was then exported and analysed with the dissociation curve software (Applied Biosystems).

pANCA and ASCA analyses

ANCA indirect immunofluorescence assay

ANCA was determined by indirect immunofluorescence (IIF) using ethanol-fixated neutrophil slides (Inova Diagnostics, San Diego, CA). Sera were incubated at a 1:40 dilution for 30 min at room temperature, washed and incubated according to the manufacturer’s instructions with fluorescein isothiocyanate-labelled rabbit antihuman IgG immunoglobulin (Inova Diagnostics, San Diego, CA). Slides were examined under UV using a Leitz Wetzler (Germany) Orthoplan microscope. Sera that exhibited fluorescence on IIF were titred to the endpoint and classified as perinuclear (pANCA) or cytoplasmic (cANCA). Interference by antinuclear antibodies, which may mimic the pANCA pattern, was ruled out by using formalin-fixed cells (Inova Diagnostic, San Diego, CA). Formalin fixation renders the neutrophil cell membrane permeable to antibodies but does not solubilise proteins such as myeloperoxidase (MPO), so the staining of MPO appears as cytoplasmic fluorescence.

ASCA ELISA assay

ASCA was measured by a standardised enzyme-linked immunosorbent assay (ELISA) using the crude mannan from S. cerevisiae uvarum I as the antigen. Coated plates with phosphopeptidomannan (PPM) extracted from yeast cells from cultures in bioreactors were obtained through Dr. D. Poulain (Centre Hospitalier Universitaire (CHU), Lille). One hundred µl of serum diluted at 1/1,000 in TNT was added. On each plate, one positive and one negative control were added, which represented, respectively, a strongly positively and negatively reacting serum against the S. cerevisiae mannan. One hundred µl of alkaline phosphatase-labelled goat anti-human IgG, IgA and IgM (heavy and light chains) diluted at 1/1,000 was used (Zymed Laboratories, Inc., San Francisco) and substrate Magia (Merck Belgolabo Belgium) for alkaline phosphatase was used to obtain a colour reaction. Reading of the plates was done at 405 nm on an ANTHOS ht II (VEL Leuven) automatic photometer. Absorbance of the individual sera was expressed relative to the absorbance of a pool of sera collected from well-characterised patients with CD. Standard reactivity curves were formed (receiver operating characteristic [ROC]) and, based on these, samples displaying an ELISA > 3.12% were considered to be positive. Investigators had no knowledge of the disease status or the diagnosis at the time of performing the pANCA or the ASCA assays.

Statistical analysis

Normality of the distribution of the data was tested by the Shapiro-Wilk test. The data are given as means ± standard deviations (SD) in the case of a Gaussian distribution and as medians with interquartile ranges (IQRs) in the case of a non-Gaussian distribution. To determine the associations between antibody responses, CARD15 or TLR4 carrier state and disease phenotype characteristics, univariate analysis was performed. All variables showing a significance of p < 0.1 in the univariate analysis were examined in a multivariate analysis using SPSS (version 14.0). Logistic regression was used (backward Wald) to identify factors associated with a more severe disease course (dependent variables were change in disease behaviour, change to fistulising disease, change to stricturing disease and the need for abdominal surgery). Odds ratios (OR) were calculated with the corresponding 95% confidence intervals (CI95%).

Genotypes were tested for the Hardy-Weinberg equilibrium and compared between different groups based on an χ2 distribution. All tests were two-tailed with predefined significance at p < 0.05.

Results

Changing phenotype with disease course

The patient characteristics are shown in Table 1.

The majority of CD children (85%) presented with traditional symptoms of diarrhoea and abdominal cramps. In addition, 37.5% also had failure to thrive and 21% had anaemia. All UC patients suffered from bloody diarrhoea at diagnosis. Seven of them (46.7%) also had failure to thrive.

Crohn’s disease

The distribution of the location at diagnosis as well as the maximal follow up is presented in Fig. 1. At diagnosis, 25% of children presented with ileal disease only, 21.2% with colonic involvement only and 53.8% with both ileal and colonic disease (Fig. 1). A large proportion presented with upper (39%) or anal (18.7%) lesions at diagnosis, and this further increased to 56% and 37.5%, respectively, during follow up. Most children (84%) presented with a non-stenosing, non-fistulating behaviour at diagnosis; however, 21.2% evolved towards stenosing and 35% towards fistulising disease after a mean of 10.7 years of follow up (Figs. 2 and 3). The disease behaviour was more susceptible to change over time than the disease location, as 36/80 (45%) patients had a change in disease behaviour compared to only 10/80 patients (12.5%) who had a change in disease location (p < 0.001).

Fig. 1
figure 1

Evolution of disease location in Crohn’s disease (CD) patients at diagnosis and at maximal follow up

Full size image
Fig. 2
figure 2

Evolution of disease behaviour in CD patients at diagnosis and at maximal follow up

Full size image
Fig. 3
figure 3

Kaplan-Meier curves in the CD patients remaining free of fistulising or stenosing complications from the moment of diagnosis

Full size image

Ulcerative colitis

The phenotypic details of the ulcerative colitis patients are also shown in Table 1. Only one patient had an extension during follow up and evolved from a left-sided colitis towards an extensive colitis.

Familial IBD

Twenty-eight CD patients (35%) and 3/15 (20%) UC patients reported to have a second relative also affected with IBD: 13 reported having a first-degree and 26 a second-degree relative affected. Seven patients even had two other members affected and one child had three other family members affected.

Among the CD patients, the median number of siblings was two and, as a median, the affected children were second in birth order (39/80 were first-born children).

Among the UC patients, the median number of siblings was three and, as a median, the affected children were second in birth order (6/15 were first-born children).

Genetic and serologic markers are associated with specific phenotypes

The results for CARD15 and TLR4 polymorphisms are summarised in Table 2. We observed a higher frequency of CARD15 variants in CD patients (55%) as compared to UC (20%, p = 0.01) and controls (21.9%, p < 0.001). Healthy relatives of CD patients more often had CARD15 variants (38.4%) than controls (p = 0.05), but less than their affected CD relatives (p = 0.007). The TLR4 299Gly minor allele frequency was 10.6% in CD, compared to 0% in UC and 5.3% in controls (p = 0.053 and 0.016, respectively).

Table 2 Allele frequencies for the NOD2 variants (Arg702Trp, Gly908Arg and Leu1007fsinsC) and the TLR4 Asp299Gly polymorphism
Full size table

The presence of CARD15 variants was associated with a more severe disease (defined as a non-inflammatory behaviour at follow up) (p = 0.004; OR 7.30; 95% CI 1.85–28.60) and with less anal involvement (p = 0.006; OR 0.086; 95% CI 0.015–0.497). The TLR4 Asp299Gly mutation in the CD patients was more frequent in patients with a younger age at diagnosis (p = 0.018; OR 0.769; 95% CI 0.619–0.956) (Fig. 5). Using linear regression, the younger the age at diagnosis, the more the patients presented with colonic disease (p = 0.021) (Figs. 4 and 5).

Fig. 4
figure 4

Proportion of patients with colonic disease as a function of the age at diagnosis

Full size image
Fig. 5
figure 5

Proportion of TLR4 Asp299Gly carriers according to the age at diagnosis

Full size image

ASCA was present in 32.9% of CD, compared to 26.7% in UC (p = 0.08) and 7.1% in controls (p = 0.001). Forty percent of UC patients were pANCA-positive (6/15), compared to 16.2% in CD and 2% in controls (all p < 0.01). In the multivariate analysis, ASCA showed inverse associations with an inflammatory non-complicated disease behaviour (p = 0.023; OR 0.26; 95% CI 0.08–0.83), with TLR4 Asp299Gly presence (p = 0.043; OR 0.104; 95% CI 0.012–0.93) and a borderline significance for less colonic involvement (p = 0.052; OR 0.169; 95% CI 0.035–1.016). No significant associations were found for the 13 pANCA-positive CD patients, although 12/13 had colonic disease (seven ileocolitis and five pure colitis) and only one pure ileitis.

Associated or concomitant (auto)immune diseases or genetic syndromes

An intriguing finding was the large number of concomitant diseases in the patients: seven patients (all CD) had ankylosing spondylitis, four suffered from psoriasis (two UC, two CD), ten from asthma (all CD), two from coeliac disease (one UC, one CD), one from type 1 diabetes (UC) and one from autoimmune thyroid disease (UC) (Fig. 6). Concomitant immune-mediated diseases were observed as frequently in CD patients (20/80, 25%) as in UC patients (3/15, 20%). In addition, two CD patients were diagnosed with Turner syndrome and one patient with Hermansky-Pudlack Syndrome.

Fig. 6
figure 6

Concomitant immune-mediated diseases in CD and UC patients (absolute numbers)

Full size image

Seven patients with concomitant diseases carried at least one CARD15 variant (35%) and this was lower than the 37/60 (61.7%) CARD15 variants in patients without concomitant diseases (p = 0.038). There was no difference in the prevalence of TLR4 variants nor ASCA or pANCA between the patients with or without concomitant diseases. Of the patients with CARD15 variants, most (5/7) had asthma, one had PSC and another had ankylosing spondylitis.

Markers of a more complicated disease course

Among the 67 CD patients presenting with inflammatory disease at diagnosis, 32 (48%) progressed to stenosing (n = 12) or penetrating (n = 20) disease over a follow up period of 10.7 years and 30 (44.8%) needed abdominal surgery. Patients who progressed to a more severe disease more often needed surgery compared to patients who remained free of stenosing or fistulising disease (27/32 or 84% versus 3/35 or 8.6%, respectively, p < 0.0001).

Patients progressing to either fistulising or stricturing disease more often had concomitant immune-mediated diseases and had a trend for more seroreactivity towards ASCA in contrast to patients who did not show a change in disease behaviour (Table 3). Patients who, during the further follow up, needed abdominal surgery more often were ASCA-positive.

Table 3 Factors associated with a complicated disease course in the subgroup of CD patients (n = 67)
Full size table

Discussion

In the present study, we evaluated the changes in disease behaviour in patients with early-onset IBD below 17 years of age. As shown in two previous studies in adults, whereas the disease location in the gut remains rather stable over time, the behaviour is susceptible to changes [11, 27, 28]. The findings in our study that 21% of CD patients developed stricturing and 35% penetrating disease after a mean follow up of 10 years almost mirror the findings in adults. This means that, with increasing disease duration, the majority of patients will develop complications, despite ongoing therapy. It seems logical that, in patients at high risk for developing complications, the early initiation of immunomodulators is necessary to change the natural course. Although there are no validated predictive factors of a complicated disease course, some clinicians have started to apply a more aggressive therapeutic approach and will introduce immunomodulators earlier in particular patients who, for instance, already present with complications at diagnosis. However, not all patients require immunomodulators or anti-TNF agents, and we must not forget that there are safety issues (risk for lymphoma, other malignancies and serious infection) associated with more aggressive medications as well. Therefore, it is crucial to identify the patients at risk who would benefit the most from these therapies.

We were particularly interested in patients who present without complications as to whether factors could be identified associated with an increased risk for complications. As expected, patients who progressed to a more severe disease in our study more often needed surgery as compared to the patients who remained free of stenosing or penetrating disease. We further found that patients who progressed more often had concomitant immune-mediated diseases and had a trend for more seroreactivity towards ASCA in contrast to patients who did not show a change in disease behaviour. Seroreactivity towards microbial antigens has been associated with more complicated disease in a number of studies [14, 28, 29, 38]. The exact mechanism behind this is not fully known, although recent data suggest that this seroreactivity is a consequence of genetic defects in receptors of the innate immune system [13].

However, we were also struck by the high prevalence of concomitant immune-mediated diseases in these patients [13, 34]. This suggests common underlying (genetic or environmental) factors [6]. It is interesting that polymorphisms in the NOD2/CARD15 gene, an intracellular receptor for bacterial muramyldipeptide, have been associated with an increased susceptibility not only to CD, but also to atopic eczema and asthma [39]. A number of overlapping regions of linkage have been demonstrated from genome-wide linkage and association studies in these diseases [35]. Therefore, given the hypothesis that CD (and also UC?) is an innate immune-deficiency syndrome, the more (genetic) defects an individual carries, the greater the risk they have for developing diseases of the so-called barrier organs, such as the skin (psoriasis, atopic eczema), the lungs (sarcoidosis, asthma), the intestine (IBD) or the joints (arthritis). It should, therefore, not come as a surprise that other diseases also are more frequent among individuals carrying CARD15 or variants in other innate immunity genes. The fact that the prevalence of CARD15 polymorphisms was not higher in the group with concomitant diseases overall points to other genes which need to be identified. At the same time, our findings suggest that these patients should be treated more aggressively from diagnosis, given their association with more complicated disease.

In conclusion, in this study, we show that more than 50% of CD patients have a disease which progresses with time and which leads to complications and the need for surgery. We identified a higher seroreactivity and the presence of concomitant immune-mediated diseases as factors associated with a greater risk for progression. Both point towards defects in PRRs as responsible factors and further underline the critical role of the innate immune system in the pathogenesis of the disease. It is also clear that CARD15 or TLR4 are not the only players.