Introduction
Ulcerative colitis (UC) is an inflammatory bowel disease (IBD) characterized by chronic inflammation of the colon and a rising incidence/prevalence worldwide in children and adolescents [
1]. Assessment of disease activity is critical for clinical management, therapeutic decisions, and monitoring of treatment targets [
2]. Endoscopic examination with tissue sampling provides a definitive assessment of the current inflammatory activity. Absence of endoscopic inflammation such as friability, blood, erosions, and ulcers in the visualized intestinal mucosa indicates mucosal healing [
3,
4]. Histologic assessment of tissue biopsies without evidence of inflammatory cell infiltrates, in particular neutrophils, indicates that histologic healing is achieved [
5,
6]. The Mayo endoscopic subscore is useful for determining mucosal healing [
7,
8], while the Nancy index is valuable for the assessment of histologic disease activity in UC [
9]. In addition to endoscopic remission, histologic remission has become an emerging treatment target in UC clinical trials [
10] and is recommended by the European Crohn’s and Colitis Organisation [
11].
Both endoscopy and histologic assessment are invasive medical procedures. Especially in pediatrics, hospitalization is often necessary to ensure an adequate (retro-/antegrade) bowel cleansing and to provide sedation or anesthesia for endoscopies and biopsies in a child-friendly environment and with sufficient quality. Bowel cleansing, analgesic sedation, and endoscopy with biopsies are not only associated with significant risks and complications for the patient, but they also tie up relevant resources of the healthcare system [
12,
13]. Therefore, non-invasive measuring tools are required to minimize the invasiveness, the discomfort, and the potential complications associated with these procedures. For pediatric UC, the Pediatric Ulcerative Colitis Activity Index (PUCAI) was developed to non-invasively predict the disease activity [
14]. It was developed using the Delphi group technique and correlated strongly with colonoscopy, Physician Global Assessment (PGA), and Mayo score [
7,
14]. However, the PUCAI was not designed to reflect histologic inflammation and therefore cannot be used to predict histologic remission/healing, which is determined as one of the therapeutic targets in clinical practice and in current studies [
15,
16]. One reason for the relevance of histologic assessment relates to the fact that patients with mucosal healing still have histologic inflammatory activity in up to 30% of cases [
17]. The absence of histologic healing is associated with an increased risk for clinical relapse, hospitalization, subsequent dysplasia, or surgery [
5,
18‐
21]. Hence, in clinical practice, endoscopy with tissue sampling is often performed before essential therapeutic decisions to include histologic inflammatory activity as an indicator of remission depth in decision-making [
22‐
24].
Accordingly, the aim of this study is to identify a minimal subset of non-invasive parameters that reflect the histologic disease activity in pediatric UC sufficiently enough and to present a feasible method how to predict histologic disease activity in daily clinical practice. To this end, we applied a Bayesian ordinal regression model and a projection-predictive feature selection, the rationale of which has already been explained in detail previously [
25].
Discussion
Our study revealed that histologic inflammatory activity in pediatric UC can be predicted parsimoniously by the combination of FC and platelet count. Other investigated laboratory parameters or subjective information obtained from medical history did not improve the prediction of histologic inflammatory activity.
For comprehensive disease management and improvement of patient outcomes, the strategy of combining treatment targets (clinical, endoscopic, and histologic remission, respectively histologic healing) to achieve a deeper level of healing is recommended in UC [
42]. In this study, we focused on the prediction of histologic healing because histologic inflammatory activity is frequently inconsistent with endoscopic inflammatory activity in approximately 30% of the cases [
17], and to date, no model has been published that provides a convenient and non-invasive prediction of histologic inflammatory activity in pediatric UC. In our study, the comparison between histologic and endoscopic inflammation scores also revealed discrepancies in the assumed severity of inflammation depending on which score is used. In fact, histo-endoscopically inactive disease is associated with reduced IBD disability as measured by patient-reported outcome measurements; however, patient-reported outcome measurements for disability and clinical disease activity cannot completely replace histo-endoscopic findings [
43]. In addition, the assessment of histologic disease activity assumes increasing relevance [
44], as histologically active disease despite endoscopic remission increases the risk of clinical relapse, hospitalization, subsequent dysplasia, or surgery [
5,
18‐
21].
Moreover, higher grades of histologic inflammatory activity were associated with a higher frequency of and a shorter time to UC progression [
45]. Recent evidence suggests that neutrophil mucosal infiltration might be a key discriminator between active disease and remission [
46‐
48] and complete resolution of neutrophil-associated acute inflammation as marker of histologic remission is of importance as a target for treatment of UC [
49]. Our model, including FC and platelet count, may serve as a proxy for histologic healing and contribute to improve the management of pediatric UC in combination with clinical findings.
Our results reveal a high predictive value for the platelet count, confirming numerous studies describing a positive correlation of platelets with inflammatory activity in UC [
50‐
52]. Our additional analyses of data from the CEDATA-GPGE registry also revealed an increasing platelet count with increasing inflammatory activity, reflected by PUCAI and PGA. All described findings confirm the crucial role of platelets in inflammation including IBD [
52‐
55]. The demonstrated relevance of platelet count as a biomarker in UC may also improve the informative value of other markers when combined with platelet count, as it has been shown in the platelet-to-lymphocyte ratio [
56] or neutrophil-to-platelet ratio [
57] in predicting disease activity. In this study, we showed that platelet count also increases with increasing histologic inflammatory activity in pediatric UC confirming its significance as a biomarker for mucosal inflammation. According to the described interference of activated platelets with leukocyte trafficking and effector functions of neutrophils and macrophages, our results underpin the predictive value of platelets, in particular for histologic inflammation, characterized by the presence of leukocytes in the mucosa.
FC reflects neutrophil migration across the inflamed gastrointestinal mucosa into the gastrointestinal tract [
58] and correlates well with endoscopic [
59] and histologic inflammatory activity in UC [
60,
61]. Complementary analysis of the CEDATA-GPGE registry emphasized the role of fecal calprotectin in assessing disease activity by confirming the positive association of FC with PUCAI and PGA.
An important requirement for the implementation of disease activity indices in clinical practice is a simple and time-saving usability in everyday clinical practice. Therefore, we demonstrated the simple application of the SHSM using FC and platelet count with a demo version of a Shiny app (
https://umrukj.shinyapps.io/shsm/). However, it should be noted that, before using it in clinical practice, our approach needs to be repeated in a prospective study with additional data to improve the reliability of the SHSM.
This study is also constrained by a limited number of patients, in particular, a low number of patients in remission. Since pediatric patients without symptoms typically underwent no endoscopy to avoid the discomfort of hospitalization as well as the potential risks and complications of endoscopy and analgesia. To improve the predictive accuracy and the reliability of the Shiny app and to validate the results, another prospective study with a larger and external cohort is required, especially including more patients in remission. For further objectification of the histologic inflammation assessment, the evaluation of biopsies using artificial intelligence may be considered to reduce inter-observer variances [
62].
Moreover, we emphasize that our methodology allows future studies to include predictors which have not yet been considered due to a high number of missing values or due to ambiguous or missing documentation (height gain, serum albumin, erythrocyte sedimentation rate, alanine-aminotransferase, gamma-glutamyltransferase, pancreas lipase, creatinine, iron metabolism parameters, height gain, and Tanner stages) or entirely newly collected parameters. Moreover, due to the proven relevance of neutrophils [
57] and lymphocytes [
56], a differential blood count should be included in further investigations. It will then become clear whether the inclusion of other predictors in the statistical model will lead to an improvement in predictive quality or whether FC and platelet count are the most meaningful parameters.
Additional statistical aspects are discussed in the electronic supplementary material.
In conclusion, we demonstrated that the combination of FC and platelet count is suitable for non-invasive prediction of histologic inflammatory activity in pediatric UC. Based on the results, this study might pave the way for the establishment of a non-invasive score to assess histologic healing in pediatric UC and to improve quality of care for children and adolescent living with IBD. The easy-to-use prediction can be performed using a Shiny app.
Acknowledgements
We thank all members of the CEDATA-GPGE study group for their support in recruitment as well as our patients and their families.
CEDATA-GPGE study group
Buderus S., Department of Pediatrics, St. Marien-Hospital-Bonn, Bonn, Germany; Bufler P., Department of Pediatrics, Charité–Berlin University Medical Center, Berlin, Germany; Däbritz J., Department of Pediatrics, Rostock University Medical Center, Rostock, Germany, Department of Pediatrics, Greifswald University Medical Center, Greifswald, Germany; Dammann S., Department of Pediatrics, Stuttgart Medical Center, Stuttgart, Germany; de Laffolie J., General Pediatrics & Neonatology, Justus-Liebig-University, Giessen, Germany; Friedt M., Department of Pediatrics, Düsseldorf University Medical Center, Düsseldorf, Germany; Hauer A., Department of Pediatrics, Graz University Medical Center, Graz, Austria; Keller K.M., Department of Pediatrics, Wiesbaden DKD Helios Hospital, Wiesbaden, Germany; Krahl A., Department of Pediatrics, Offenbach Sana Medical Center, Offenbach, Germany; Laaß M., Department of Pediatrics, Dresden University Medical Center Carl Gustav Carus, Dresden, Germany; Lang T., Medical Center St. Hedwig, KUNO University Children’s Clinic, Regensburg, Germany; Posovszky C., Gastroenterology & Nutrition, University Children’s Hospital Zurich, Zurich, Switzerland; Rodeck B., Department of Pediatrics, Osnabrück Christian Children’s Hospital, Osnabrück, Germany; Trenkel S., Department of Pediatrics, Western Brandenburg Medical Center, Potsdam, Germany.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.