What Is Known/What Is New
What Is Known
- Ileal nonintubation occurs in 20% to 25% of colonoscopies in pediatric Crohn disease.
- In adults, magnetic resonance enterography including enema and MaRIA score calculation highly agrees with ileocolonoscopy regarding mucosal inflammation.
- MaRIA score has not been used in pediatric studies.
What Is New
- There is a moderate agreement between the simple endoscopic score of Crohn disease and magnetic resonance enterography in pediatric Crohn disease.
- Calculation of the ileal MaRIA score can be used to impute ileal simple endoscopic score of Crohn disease in patients without ileal intubation.
- Inclusion of predicted simple endoscopic score of Crohn disease reveals additional patients with disease, who would have been considered normal otherwise.
Mucosal healing (MH) has been widely recognized as an important outcome of treatment in inflammatory bowel diseases (IBDs). Consequently, regulators require including measures of mucosal inflammation in clinical trials. Ileocolonoscopic evaluation is considered the most accurate method for reflecting intestinal inflammation, but it assesses only part of the bowel and does not provide data on inflammation beyond the mucosa. The adjuvant use of magnetic resonance enterography (MRE) can address these limitations. Moreover, although the cecum has been reached in >90% of colonoscopies recorded in the large multicenter pediatric Eurokids registry (n = 1227) (1), the ileum was not intubated in 26% of cases, leading to inability to calculate the Simple Endoscopic Score of Crohn Disease (SES-CD) score (1).
Because colonoscopy is becoming the primary outcome of clinical trials, a large percentage of children will have missing data from the most relevant disease location, biasing the results and lowering the sample size. The SES-CD will be zero in a child with isolated ileitis, if the colonoscopy reaches the cecum only. In these cases it may be appealing to impute the SES-CD by results obtained from MRE, but no data exist to base these estimation.
We therefore aimed to phenotype the children without ileal intubation from a large prospective multicenter cohort of pediatric CD, and to explore the ability to impute the ileal SES-CD score from MRE when needed.
This is an ancillary analysis of the prospective ImageKids study (clinicaltrials.gov NCT01881490) which aimed to construct 2 new MRE-based scoring systems for pediatric CD, one of inflammation (the Pediatric Inflammatory Crohn Disease MRE Index [PICMI]) and one of damage (the pediatric MRE Damage Index in Crohn Disease). A total of 240 children (<18 years old) with CD have been enrolled in this study at 22 pediatric IBD centers in North America, Europe, Australia, and Israel.
Children with known or newly diagnosed CD according to accepted criteria (2) were recruited and underwent a baseline ileocolonoscopy, followed by an MRE examination within 14 days without a change in treatment. The study has been approved by the Ethics Committees of all centers, and all patients provided written informed consent.
Description of Variables
Explicit data were recorded regarding clinical and disease characteristics, prior therapies, and laboratory tests. Clinical disease activity was scored using a gastroenterologist global assessment of disease activity and the weighted pediatric Crohn disease activity index (3). The first 120 children were followed through 18 months when MRE examination was repeated; the present study includes data from the baseline visit only to avoid repeated measures bias.
The SES-CD items were scored during ileocolonoscopy at each segment (rectum, sigmoid, descending, transverse and ascending colon, and terminal ileum) from 0 to 3 (4) (Appendix 1, Supplemental Digital Content 1, http://links.lww.com/MPG/B660). MRE sequences and their scoring system were standardized across centers (Appendix 2, Supplemental Digital Content 2, http://links.lww.com/MPG/B661). Each protocol included localizer sequences, a motility sequence in the coronal plane, followed by a series of coronal and/or axial rapid T2-weighted, diffusion-weighted and T1-weighted gradient echo sequences. T1-weighted sequences were performed pre- and postintravenous gadolinium. An intravenous antispasmodic agent (glucagon or hyoscine butylbromide) was administered following the motility sequence. The radiologist at each enrolling site, blinded to the ileocolonoscopy results, prospectively completed a standardized MRE report. Central reading was performed by 2 independent radiologists, and the final report consisted of the average of both. Severity of inflammatory disease activity on MRE was scored on 100 mm visual analogue scales (VAS) for each bowel section and globally for the entire bowel by 2 independent radiologists (ranging from “remission” to “fulminant” and calculating the average score). Scoring the MRE global assessment for inflammation has been previously validated on our cohort (5) and relies on bowel wall thickness, T1 enhancement, T2 hyperintensity, diffusion-weighted imaging signal, mucosal signs of ulcerations, and mesenterial signs (such as combs sign and lymph node enlargement). A VAS score <20 mm has been previously validated against the MaRIA to reflect MH (6).
Data are presented as means ± standard deviations or medians (interquartile range [IQR]) for continuous variables and as frequencies and percentages for categorical variables. The relationships of MRE and weighted pediatric Crohn's disease activity index with SES-CD were assessed by the Spearman's correlations. Independent Student t tests were performed to compare SES-CD components and MRE-VAS in patients with and without ileal intubation. Linear regression analysis was used to estimate the ileum SES-CD by the MaRIA score. Data were analyzed with SPSS software version 23.0 (SPSS Inc, Chicago, IL). The significance levels were set at 0.05.
A total of 237 children of ages 1 to 17.5 years were included in the study (in 3/240 endoscopic scoring was missing). The ileum was not intubated in 40 children (17%), of whom the right colon was reached in 38 (95%) (Table 1). SES-CD score was available in the right colon in 231 of 237, transverse colon in 231 of 237, left colon in 236 of 237 and rectum in all 237 patients. We were able to calculate the MaRIA score in 175 of 237 (74%) children; the others had missing MRE components required for the calculation of the index (mainly the specific enhancement component) (7).
Association of Magnetic Resonance Enterography and Simple Endoscopic Score of Crohn's Disease in General
The correlation between SES-CD and MRE-VAS inflammation score was moderate to high (ileum r = 0.536, ascending colon r = 0.450, transverse colon r = 0.489, descending colon r = 0.553, sigmoid r = 0.658, rectum r = 0.632, and total scores r = 0.658; all P < 0.001). The percentage agreement between SES-CD and the MRE-VAS inflammation scores for reflecting MH was also moderate-high across the ileum and colonic segments (70%–85%; all P < 0.001) (Table 2). The sensitivity and specificity of the MRE-VAS for reflecting MH (ie, SES-CD <3) were 92% (95% confidence interval [CI] 0.84–0.96) and 53% (95% CI 0.43–0.63), respectively. There was a moderate correlation between SES-CD and the MaRIA subscores of the ileum (r = 0.418 [P < 0.01]) and the agreement for reflecting MH was fair (67%, k = 0.264, P = 0.006). The sensitivity and specificity of the ileal MaRIA for MH was 41.4% (95% CI 0.74–0.91) and 83.5% (95% CI 0.29–0.55), respectively. The agreement between SES-CD and MRE-VAS and the correlation between SES-CD and the ileal MaRIA subscore were similar in patients younger than 10 years and older patients.
Phenotyping the Children Without Ileal Intubation
The demographic characteristics of the patients without ileal intubation were similar to those with completed ileocolonoscopy; however, those without ileal intubation had higher C-reactive protein (median 5.0 [IQR 1.3–18.8] vs 12.6 [4.9–36], respectively, P = 0.002), and a higher rate of enteral nutrition therapy (17/197 [9%] vs 10/40 [25%], respectively, P = 0.013; Table 1). There was no difference in fecal calprotectin levels between those with or without ileal intubation (611 [148–1279] vs 564 [221–1224], respectively, P = 0.95). No difference was found in SES-CD, MRE-VAS of inflammation, and ileal MaRIA scores between the patients receiving or not receiving enteral nutrition (MaRIA 14.9 ± 10.9 vs 17.3 ± 10.3; MRE-VAS 17.1 ± 24.1 vs 19.0 ± 27.1; SES-CD 3.04 ± 3.5 vs 3.8 ± 3.4, all NS). The median SES-SD of the right colon was higher in patients without ileal intubation (3 [IQR 0–5]) compared with the others (0 [0–3]; P = 0.021) (Fig. 1). Similarly, the mean ileal MaRIA subscore was higher in children without ileal intubation (20.5 ± 7.1 vs 15.1 ± 10.8; P = 0.0018). Among the 40 children without ileal intubation, 16 (40%) had evidence of ileal disease on MRE (ie, VAS >20 mm), in light of normal colonoscopic evaluation (colonic SES-CD <3). These findings indicate that isolated active ileal disease may be missed in 7% (16/237) of children and labeled as having MH when considering SES-CD only as an outcome without MRE.
Predicting Ileal Simple Endoscopic Score of Crohn's Disease From Magnetic Resonance Enterography
The following model was constructed for predicting the ileal SES-CD subscore from MRE-VAS of inflammation of that segment: SES-CDileum = 1.986 + 0.076X (MRE-VASileum) and rounding to the nearest whole number (R = 0.278). The mean imputed ileal SES-CD subcore of the 40 children without ileal intubation using this equation (3.7 ± 2.2) was similar to the other 197 children with ileal intubation (3.2 ± 1.9; P = 0.172). In 7 of the 16 children (44%) without ileal intubation and normal colonoscopy to the cecum, the SES-CD changed to 5 to 8 (mild-moderate inflammation) after imputing the predicted ileal SES-CD score from the MRE-VAS.
Since the MRE-VAS is subjective, we repeated the prediction model from the MaRIA subscore of the ileum, available in 178 of the 237 patients, of whom 33 without ileal intubation. When the MaRIA score of the ileum was zero we considered the SES-CD to be zero as well; otherwise the following model applied: SES-CDileum = 1.145 + 0.169 × MaRIAileum rounded to the nearest whole number (R2 = 0.17).
To explore the accuracy of the model, we first verified that the proportion of patients with endoscopic ileal disease (ie, SES-CD ≥3) is comparable to the MaRIA estimation in the 145 children with both measures. The mean ileal SES-CD was similar whether imputed by MaRIA (3.89 ± 3.64) or obtained directly by ileocolonoscopy (3.89 ± 1.91). By imputing the SES-CD score from the MaRIA score, 88/145 (61%) had ileal SES-CD ≥3 as compared with 85 of 145 (59%) obtained directly by ileocolonoscopy (P > 0.2). The actual per-patient agreement was apparent in 144 children (75%). Applying the model to the 33 patients without ileal intubation, revealed that 29 of 33 (88%) had ileal disease and 8 (28%) had a normal colonoscopy to the cecum but an imputed ileal SES-CD ≥3. This means they would have been falsely characterized as normal based on colonoscopy only without the addition of the MRE.
To the best of our knowledge, this study is the first attempt to impute missing SES-CD ileal scores from concurrent MRE. It is also by far the largest study to correlate MRE with the SES-CD in children. The ImageKids study is a prospective study performed in major international pediatric IBD centers that are also typically chosen to be included in regulatory clinical trials. It is therefore concerning that the ileum has not been intubated in 17% of children, a figure that stands in the middle of the reported range in multicenter pediatric cohorts—between 24% in the large Eurokids registry to 7% in the RISK cohort (personal communication, J.H.). Undoubtedly, failing to intubate the ileum poses a significant challenge when using the SES-CD as a primary outcome in clinical trials. Particularly since the SES-CD data would not be missing at random, we found that the ileal intubation rate is associated with the degree of inflammation in the ascending colon and ileum. Some of those would have falsely classified as having MH if considering the SES-CD score to the cecum, and the efficacy of trial medication may be wrongly interpreted in such cases. We thus propose an equation for predicting the missing ileal SES-CD subscore from either the MaRIA or radiological global assessment, which added 75% to the accuracy of the SES-CD when the colonoscopy was incomplete. This imputation may not be perfect but the alternative of assigning a score of 0 to an unseen ileum seems more biased.
In our study the correlation of the MaRIA score with the SES-CD was r = 0.658, slightly lower than found in adults (r = 0.71) (8,9), possibly since we performed the MRE without an enema to enhance the accuracy of the colonic score. MRE is not tolerated well by many children and the addition of an enema further reduces its feasibility. Stoppino et al (9), in a study of adults treated with anti-tumor necrosis factor for CD, demonstrated a high accuracy of predicting MH by the MaRIA using a cutoff point of 30.8 (SES-CD ≤2, AUC 0.967, sensitivity 93%, specificity 77%). In another prospective study of 48 adults, MaRIA score of <11 provided 90% accuracy, and a score of <7 provided 83% accuracy, for detecting endoscopic ulcer healing, and MH, respectively (10). MaRIA score has been rarely used in pediatric studies, and currently the development of the PICMI is underway via the ImageKids study (11). The PICMI is aimed specifically for children and thus will avoid using enema and Gadolinium agent, which has been recently detected in the brain of patients who underwent repeated magnetic resonance imaging prompting the Food and Drug Administration to issue a safety warning.
Several small (<50 children), mainly retrospective, pediatric studies correlated MRE with endoscopic findings in children with CD. The sensitivity and specificity of MRE to predict histological inflammation has been found by Campari et al (12) (n = 37 children) and Barber et al (13) (n = 15 children), to be 94%/64%, and 90%/60%. Pomerri et al (14) (n = 24 children) reported a sensitivity and specificity of MRE for active endoscopic disease of 89% and 100%, respectively, and a moderate-strong correlation (r = 0.70, P = 0.001) of the global MaRIA score with SES-CD. We found a similar sensitivity and specificity of the MRE-VAS for MH (92% and 53%, respectively) to those reported by Campari and Barber. Similarly to Pomerri, we found a moderate correlation between SES-CD and MaRIA score; however, only the MaRIA subscores of the ileum was performed (r = 0.418 P < 0.01).
A recent systematic review and meta-analysis of the performance of MRE in pediatric IBD included 18 studies (9 prospective, 9 retrospective, all <100 included children) involving a total of 687 patients (15). Despite substantial heterogeneity across the studies, MRE demonstrated a high diagnostic performance for active inflammation in children with known or suspected IBD. The pooled sensitivity to accurately detect inflammation was 83% (95% CI 75%–89%), and specificity of 93% (90%–95%), and in a per-patient analysis 86% and 91%, respectively. Our results, based on a large prospective cohort of 237 children (more than double the largest pediatric study to date on MRE in CD) and using strict standardized methods, show a sensitivity of 92% (95% CI 0.84%–0.96%) and a specificity of 53% (0.43%–0.63%) of the MRE-VAS to detect MH. We found a moderate-high correlation and a moderate agreement (highest in the ileum, k = 0.508) between MRE-VAS of inflammation and SES-CD and a moderate correlation between MaRIA and SES-CD scores.
This study has several strengths including its large sample size, the prospective design and evaluation, the muticenter nature, and rigorous standardization of MRE sequences and procedures across all sites but it is not without limitations. MaRIA score was unavailable in a quarter of the patients given missing MRE sequences required for calculating the score. In these children we supplemented the evaluation with radiologic global assessment of inflammation but this is a subjective measure. The presence of an expert magnetic resonance imaging radiologist specializing in gastroenterology is important and sufficient to reproduce our data, and central reading is not mandatory, although often used in clinical multicenter studies. Lastly, the MRE performed in the ImageKids study does not include an enema, as traditionally required for calculating the MaRIA score in adults. Adding an enema to a pediatric MRE would significantly jeopardize feasibility. Therefore, the development of a pediatric MRE scoring system, the PICMI, which does not include an enema and does not require gadolinium is underway, based on the same study population.
MRE should be included as a secondary outcome in clinical trials of pediatric CD, whereas colonoscopy remains the criterion standard. The goal of including MRE is to supplement the data obtained from ileocolonoscopic evaluation, not only for imputing missing SES-CD data as shown here, but also for assessing the entire width of the bowel wall (rather merely the mucosa) and the degree of inflammation in areas outreaching the colonoscope. This is particularly important in children who have panenteric disease much more often than in adults (16).
1. De Bie CI, Buderus S, Sandhu BK, et al. Diagnostic workup of paediatric patients with inflammatory bowel disease in Europe: results of a 5-year audit of the EUROKIDS registry. J Pediatr Gastroenterol Nutr
2. Levine A, Koletzko S, Turner D, et al. ESPGHAN revised porto criteria for the diagnosis of inflammatory bowel disease in children
and adolescents. J Pediatr Gastroenterol Nutr
3. Turner D, Griffiths AM, Walters TD, et al. Mathematical weighting of the pediatric Crohn's disease activity index (PCDAI) and comparison with its other short versions. Inflamm Bowel Dis
4. Daperno M, D’Haens G, Van Assche G, et al. Development and validation of a new, simplified endoscopic activity score for Crohn's disease: the SES-CD. Gastrointest Endosc
5. Focht GTT, Church P, Walters TD, et al. Damage and inflammatory activity in pediatric Crohn's disease (CD) based on radiologist and gastroenterologist physician global assessment. J Crohn Colitis
6. Weinstein-Nakar I, Focht G, Church P, et al. Associations among mucosal and transmural healing and fecal level of calprotectin in children
with Crohn's disease. Clin Gastroenterol Hepatol
7. Rimola J, Ordas I, Rodriguez S, et al. Magnetic resonance imaging for evaluation of Crohn's disease: validation of parameters of severity and quantitative index of activity. Inflamm Bowel Dis
8. Coimbra AJ, Rimola J, O’Byrne S, et al. Magnetic resonance enterography
is feasible and reliable in multicenter clinical trials in patients with Crohn's disease, and may help select subjects with active inflammation. Aliment Pharmacol Ther
9. Stoppino LP, Della Valle N, Rizzi S, et al. Magnetic resonance enterography
changes after antibody to tumor necrosis factor (anti-TNF) alpha therapy in Crohn's disease: correlation with SES-CD and clinical-biological markers. BMC Med Imag
10. Ordas I, Rimola J, Rodriguez S, et al. Accuracy of magnetic resonance enterography
in assessing response to therapy and mucosal healing in patients with Crohn's disease. Gastroenterology
11. Turner D, Gavish M, Focht G, et al. Development of the Pediatric Inflammatory Crohn's MRE Index (PICMI): results from the ImageKids Study. ECCO Annual Meeting
, February 16 2018, Vienna. 2018. Abst P128.
12. Campari A, Napolitano M, Zuin G, et al. Colonic inflammation in pediatric inflammatory bowel disease: detection with magnetic resonance enterography
. Pediatr Radiol
13. Barber JL, Lozinsky AC, Kiparissi F, et al. Detecting inflammation in the unprepared pediatric colon—how reliable is magnetic resonance enterography
? Pediatr Radiol
14. Pomerri F, Al Bunni F, Zuliani M, et al. Assessing pediatric ileocolonic Crohn's disease activity based on global MR enterography scores. Eur Radiol
15. Yoon HM, Suh CH, Kim JR, et al. Diagnostic performance of magnetic resonance enterography
for detection of active inflammation in children
and adolescents with inflammatory bowel disease: a systematic review and diagnostic meta-analysis. JAMA Pediatr
16. Van Limbergen J, Russell RK, Drummond HE, et al. Definition of phenotypic characteristics of childhood-onset inflammatory bowel disease. Gastroenterology