Serological cytokine signature in paediatric patients with inflammatory bowel disease impacts diagnosis

Endoscopy is a central tool for diagnosing and evaluating paediatric inflammatory bowel diseases (PIBD), but is too invasive to be frequently repeated in young children. Furthermore, it is challenging to distinguish Crohn’s disease (CD) from ulcerative colitis (UC) endoscopically. This study aimed to determine biomarkers useful for the diagnosis of PIBD. Cytokines, chemokines, and growth factors were quantified in the sera of 15 patients with CD or UC, at disease onset prior to treatment, and 26 age-matched controls. Correlation of cytokine levels with the paediatric CD activity index (PCDAI) and the paediatric UC activity index (PUCAI) was analysed. Interleukin (IL)-6, IL-13, IL-7, and vascular endothelial growth factor were higher in the CD group than in the UC group. The receiver operating characteristic curve analysis showed that IL-7 was a putative biomarker for distinguishing CD from UC (area under the curve: 0.94). Granulocyte–macrophage colony-stimulating factor was associated with PCDAI, and an IL-1 receptor antagonist, IL-6, and macrophage inflammatory protein-1β were associated with PUCAI. These findings indicate significant differences in cytokine signatures among patients with new-onset PIBD, which may improve accuracy in diagnosing PIBD.


Discussion
Our results indicated that cytokines were differentially expressed in the sera of CD and UC patients prior to treatment, suggesting that these biomarkers can aid in the diagnosis of PIBD. We observed 12 differentially expressed cytokines in the new CD cases, but none in the new UC cases when compared to controls. Furthermore, we  www.nature.com/scientificreports/ identified IL-7 as a candidate biomarker for distinguishing CD from UC. There were no factors associated with endoscopic scoring in CD, whereas 4 factors were associated with UC-namely, IL-13, IL-7, MCP-1, and VEGF. GM-CSF, IL-1ra, IL-6, and MIP-1β were correlated with PCDAI in CD and PUCAI in UC. A previous study attempting to identify serum cytokine profiles to differentiate CD and UC in adult patients reported that IFN-γ, IL-6, and IL-7 were higher in CD and growth-regulated oncogene and TNF-α were higher in UC than in controls, but no cytokines were differentially expressed between CD and UC 24 . The study used a generalised linear model to discriminate UC and CD, including age, sex, and 15 cytokines as parameters and observed an AUC of 0.936 24 . The current study observed clear differences in cytokine profiles between CD and UC, with IL-7 alone having the highest diagnostic accuracy. The discrepancy in results between the previous and current studies might be attributed to the age and treatment of participants. As mentioned earlier, PIBD has different clinical presentation than adult IBD, which suggest differences in underlying pathology. The distinct cytokine expression profiles observed in this study, especially in CD patients, likely reflect the differences in pathology between adults and children. Table 2. Concentrations (pg/mL) of cytokines, chemokines, and growth factors in the serum of CD, UC, and control patients. Data are presented as median or numbers (proportions). Differences between the three groups were statistically compared using the Kruskal-Wallis test (*). P-value shows the results of Mann-Whitney U test with Bonferroni correction ( †). ns not significant, CD Crohn's disease, UC ulcerative colitis, IFN interferon, IL interleukin, TNF tumour necrosis factor, ra receptor antagonist, IP-10 IFN-γ-induced protein 10, MCP monocyte chemoattractant protein, MIP macrophage inflammatory protein, G-CSF granulocyte-colonystimulating factor, GM-CSF granulocyte-macrophage colony-stimulating factor, VEGF vascular endothelial growth factor, FGF fibroblast growth factor. www.nature.com/scientificreports/ The use of medication was also different between our current study and others. No participants in this study received IBD treatment, such as 5-aminosalicylic acid, steroids, immunomodulators, biologics, antibiotics, blood cell apheresis, or surgery, prior to sample collection. These treatments can decrease gastrointestinal inflammation and affect serum cytokine expression profiles. It is vital to determine baseline cytokine profiles of IBD patients to determine their utility as diagnostic biomarkers, but most studies on serum cytokines in PIBD do not indicate whether or not participants underwent treatment. Fujitake et al. investigated IL-4, IL-5, IL-6, IFN-γ, TNF-α, and TGF-β1 in CD patients who were not taking medication and reported that serum IL-6 and TNF-α were   www.nature.com/scientificreports/ higher in the acute phase than in the remission phase 20 . Kader et al. reported that the placental growth factor, IL-7, TGF-β1, and IL-12p40, were higher in the active phase of CD than in the remission phase 25 . We found 12 cytokines that were increased in the active CD at disease onset. We did not compare the active and remission phases, but rather diseases and controls. The larger number of differentially expressed cytokines in this study may be attributable to this difference in comparison. IL-7 had the highest diagnostic accuracy to distinguish CD from UC. IL-7 was shown to be required for T-cell development and is secreted from stromal and epithelial cells 26 . It is produced by intestinal epithelial cells, and its overexpression induces spontaneous development of chronic colitis in mice 27 . Furthermore, the IL-7 receptor has been proposed as a therapeutic target for IBDs 28,29 . These previous studies strongly demonstrate that IL-7 is involved in the pathology of IBD. However, there is a report showing a contradictory result, in which serum IL-7 levels were lower among patients at diagnosis and in the remission and active phases than among controls 30 . These results indicate that serum IL-7 levels are likely dependent on the patient's condition. We found that IL-7 was significantly higher in new paediatric cases of CD than in those of UC. Further studies are required to elucidate whether IL-7 can serve as a biomarker to distinguish CD from UC in PIBD patients.
We found that CRP and ESR were significantly higher in the CD group than in the UC group. Consistently, Fagan et al. 31 reported elevation of CRP and ESR in both CD and UC patients, with higher elevation in CD than in UC for all categories of disease severity. Interestingly, Torres et al. recently reported that serum antibodies and proteins change 5 years before CD, but not at UC onset in adult patients 32 . Although this previous study differs from ours in terms of the age of participants and timing of measurement, they both indicate that inflammation in CD is intensive and distinct from that in UC. Collectively, our results indicated that more cytokines were upregulated in CD than in UC, and that the extent of systemic inflammation might be higher in CD than in UC among PIBD patients.
A recent study of children with CD demonstrated the lack of correlation among the clinical disease activity index, the PCDAI, and endoscopic severity 33 . Consistent with this, we did not observe a significant correlation Table 4. Correlation between the endoscopic scores and serum cytokine levels. CD Crohn's disease, SES-CD simple endoscopic score for CD, UC ulcerative colitis, UCEIS ulcerative colitis endoscopic index of severity, IFN interferon, IL interleukin, TNF tumour necrosis factor, ra receptor antagonist, IP-10 IFN-γ-induced protein 10, MCP monocyte chemoattractant protein, MIP macrophage inflammatory protein, G-CSF granulocyte-colony-stimulating factor, GM-CSF granulocyte-macrophage colony-stimulating factor, VEGF vascular endothelial growth factor, FGF fibroblast growth factor.  34 . In accordance with this, there was no significant correlation between PUCAI and UCEIS in this study. We found that GM-CSF, a potent growth and differentiation factor in myeloid cells, was correlated with PCDAI. Myeloid cells are responsible for the anti-microbial activity, and neutralisation of GM-CSF using anti-GM-CSF autoantibodies is proposed to be associated with disease progression and relapse of IBD 16,17 . These results suggest that GM-CSF may be a potential marker of disease activity in new-onset paediatric CD patients. PUCAI was correlated with a model including IL-1ra, MIP-1β, and IL-6, although serum levels of these cytokines were not significantly different between the UC group and controls. This indicates that these factors may reflect disease activity, but not morbidity. IL-6 has been reported to be associated with steroid resistance and disease activity in paediatric patients with severe UC 18 and may be a putative marker of disease activity in paediatric UC patients. In adult patients, reduction of serum IL-6 measured within a couple of months after beginning therapy with biologics can be used to predict the effects of the biologics 12 months after their use to treat both CD and UC, indicating that IL-6 levels correlate with disease activity in both children and adults 35 .

Spearman's r p-value Spearman's r p-value
A limitation of this study was the small number of subjects from a single institution. An increase in the number of patients would result in more uniform groups and potentially reveal differences among PIBD subtypes with different immunopathogeneses. Further, the extent of inflammation differed between UC and CD in this study.  www.nature.com/scientificreports/ Because it was unclear how this difference affects the serum cytokine levels, we tested the correlation of cytokine levels with CRP or ESR (see Supplementary Fig. S1 online). There was a moderate correlation between CRP and each of IL-6 and VEGF, supporting the specific increase in IL-7 in CD patients in the current study. However, the number of participants was small, and thus, future studies should investigate whether and how active inflammation in CD with high ESR and CRP levels affect serum cytokine levels. The current results thus need to be confirmed in a larger patient population from multiple institutes before they can be applied in clinical settings. In conclusion, endoscopic or pathological findings in new-onset PIBD are different from those in adults. Rothschild et al. 36 reported that granulomas were only found in 34% of paediatric CD patients. Additionally, crypt abscesses, which are the basis for the diagnosis of UC, are a nonspecific manifestation of several forms of enteritis. An epidemiological survey estimated that the morbidity rate of IBDU was approximately 15% in children and 5% in adults 37 . Therefore, easily accessible biomarkers for diagnosing new-onset PIBD are clinically relevant. The serological features of several cytokines allow for disease identification. Finally, we found that serum IL-7 was a useful non-invasive marker for distinguishing CD from UC in new cases of PIBD.

Methods
Study population and sample collection. We enrolled consecutive patients who had been diagnosed with IBD, either CD or UC based on Porto criteria 38 , at the Department of Pediatrics of Gunma University Hospital or Saitama Children's Medical Center in Japan from June 2008 to October 2015. Consequently, CD and UC diagnoses were established based on standard clinical, endoscopic, and histological criteria (Table S1). Endoscopy was performed at the same time of serum collection. The inclusion criteria for patients included: active stage of disease, but no prior treatment at the time of serum sampling. Exclusion criteria included: IBDU diagnosis, history of autoimmune diseases, or previous treatment (5-aminosalicylic acid, steroids, immunomodulators, biologics, antibiotics, blood cell apheresis, or surgery) at the time of serum sampling. Control samples were obtained from age-matched patients with chronic functional constipation but without IBD or any other inflammatory disorders.
Collection of clinical information. Clinical characteristics including age at diagnosis, sex, laboratory findings (haemoglobin, CRP, ESR, white blood cell count, thrombocytes, fibrin/fibrinogen degradation products, D-dimer, and serum albumin), disease types, clinical activity, endoscopic findings, and treatment received were obtained from medical records. Disease location was based on the Paris classification: L1 (terminal ileum), L2 (colonic), L3 (ileocolonic), and L4 (upper disease) for CD and E1 (ulcerative proctitis), E2 (left-sided UC), E3 (extensive), and E4 (pancolitis) for UC 15,39 . Disease activity was assessed using the PCDAI 40 or PUCAI 41 . The endoscopic index was graded according to SES-CD 22 or UCEIS 23 . The SES-CD assesses mucosal ulcer size, ulcerated surface, endoscopic extension, and presence of stenosis 22  Serum samples were frozen at − 80 °C immediately after collection and were stored until analysis. All samples were thawed on ice, vortexed, centrifuged at 14,000 × g for 10 min at 4 °C, and diluted with the standard diluent in the kit (Bio-Plex Pro Human Cytokine 27-Plex Assay #M500KCAF0Y; ratio 1:4 by volume). Each sample was measured more than twice. The mean values of measurements were used as the representative values for each subject.
Statistical analyses. Patient characteristics are presented as median and interquartile range or as proportions. Differences between groups were determined using the Mann-Whitney U test, Student's t-test, or Fisher's exact test. Serum cytokine concentrations are presented as mean ± SEM from three independent measurements. Differences between the three groups (CD, UC, and control groups) were statistically compared using the Kruskal-Wallis test, and post hoc Mann-Whitney U test with Bonferroni correction was used for comparisons between two of the three groups. A ROC curve was used to determine the appropriate cut-off cytokine levels for distinguishing CD from UC. Correlations between the endoscopic scores (SES-CD, UCEIS) and serum cytokine levels were established using nonparametric Spearman's correlation. Multiple linear regression analysis was used to predict disease activity (PCDAI, PUCAI) based on cytokines.
Statistical analyses were performed using JMP14 (SAS Institute Inc., Cary, NC, USA) and GraphPad Prism version 7.00 for Windows (GraphPad Software, La Jolla, CA, USA), and p-values < 0.05 were considered statistically significant.
Ethical considerations. This study was approved by the Medical Ethical Committee of the Gunma University Hospital (protocol number HS2019-018) and Saitama Children's Medical Center (protocol number 2019-05-024) and was conducted in accordance with all ethical principles of the Seventh Revision of the Helsinki Declaration from 2013. Samples were initially collected for another study, namely "Epigenetics of Inflammatory Bowel Disease" (protocol number 107: approved by the Medical Ethical Committee of the Gunma University Hospital). Serum samples were obtained after patients, and their parents/guardians provided informed consent, Scientific RepoRtS | (2020) 10:14638 | https://doi.org/10.1038/s41598-020-71503-y www.nature.com/scientificreports/ and participation in the study was voluntary. All experiments were performed in accordance with the approved guidelines.

Data availability
The datasets generated and analysed during this study are available from the corresponding author upon reasonable request.