Bedside Intestinal Ultrasound Predicts Disease Severity and the Disease Distribution of Pediatric Patients With Inflammatory Bowel Disease: A Pilot Cross-sectional Study

Abstract

Background

Intestinal ultrasound (IUS) is a noninvasive tool to assess bowel inflammation. There is a paucity of data on its accuracy in pediatric patients.

Aim

The aim of this study is to evaluate the diagnostic performance of bowel wall thickness (BWT) measured using IUS compared with endoscopic disease activity in children suspected of having inflammatory bowel disease (IBD).

Methods

We conducted a single-center cross-sectional pilot study of pediatric patients suspected to have previously undiagnosed IBD. Endoscopic inflammation was graded using segmental scores of the Simple Endoscopic Score for Crohn's Disease (SES-CD) and the Ulcerative Colitis Endoscopic Index of Severity (UCEIS) and classified as having healthy, mild, or moderate/severe disease activity. Association between BWT and endoscopic severity was assessed using the Kruskal-Wallis test. The diagnostic performance of BWT to detect active disease at endoscopy was evaluated using the area under the receiver operating characteristic curve; sensitivity and specificity were calculated.

Results

In all, 174 bowel segments in 33 children were assessed by IUS and ileocolonoscopy. An elevated median BWT was associated with increased bowel segment disease severity, classified by the SES-CD (P < .001) and the UCEIS (P < .01). Using a cutoff value of 1.9 mm, we found that the BWT had an area under the receiver operating characteristic curve of 0.743 (95% CI, 0.67-0.82), a sensitivity of 64% (95% CI, 53%-73%), and a specificity of 76% (95% CI, 65%-85%) to detect inflamed bowel.

Conclusion

Increasing BWT is associated with increasing endoscopic activity in pediatric IBD. Our study suggests that the optimal BWT cutoff value for detecting active disease may be less than that seen in adults. Additional pediatric studies are needed.

Key Messages.

• Bowel wall thickness (BWT), as measured on intestinal ultrasound (IUS), is a reliable indicator of disease activity in the adult patient population.

• Increasing BWT is associated with increased intestinal inflammation, as seen on ileocolonoscopy, including when using endoscopic severity scores. The BWT cutoff for detecting active inflammation in pediatric patients may be lower than that reported for adult IBD patients.

• This study highlights that the characteristics of inflammatory findings on IUS differ in children compared with the reported adult standards.

Introduction

Inflammatory bowel diseases (IBDs) are autoimmune disorders characterized by waxing, waning, and often progressive course. The approach to managing these conditions has changed dramatically in the last 2 decades, with an emphasis on reaching multiple treatment targets such as normalization of clinical symptoms and biochemical markers (eg, C-reactive protein [CRP] and fecal calprotectin), depending on time from treatment initiation.¹ Ultimately, reaching mucosal healing has been associated with prolonged sustained remission and lower complication rates over time.^2,3 However, objective repeated assessments of inflammation throughout treatment are necessary to achieve this goal.⁴

The reference standard for assessing mucosal healing in pediatric patients with IBD remains to perform an upper and lower gastrointestinal tract endoscopy, often graded using an endoscopic scoring system such as the Simple Endoscopic Scoring system for Crohn’s Disease (SES-CD) or the Mayo Endoscopic Score (MES) for ulcerative colitis (UC).^5,6 The Ulcerative Colitis Endoscopic Index of Severity (UCEIS) is a validated score used to grade endoscopic activity and used in clinical trials.⁷ Magnetic resonance enterography (MRE) and capsule endoscopy are the standard modalities for assessing small intestinal disease activity. Additionally, there is growing evidence that transmural healing, which may be evaluated by cross-sectional imaging modalities such as MRE or CT enterography (CTE), should be the treatment target.^8,9 Endoscopic assessments are invasive, resource-intensive procedures that require general anesthesia for most children. Magnetic resonance enterography is expensive, with limited availability in some regions.¹⁰ In addition, the intake of large volumes of oral contrast agent and the need for anesthesia in young children to minimize the risk of motion artifact and to provide necessary breath-holding limit the use of this seemingly less invasive modality. Finally, endoscopy, MRE, and even capsule endoscopy are time-consuming modalities requiring additional lengthy appointments at the hospital.

Intestinal ultrasound (IUS) is a promising and appealing diagnostic and monitoring tool for pediatric IBD. When available at the bedside, this modality can be used during an initial consultation visit to assess baseline disease activity and during follow-up visits to evaluate ongoing bowel wall inflammation. Intestinal ultrasound is noninvasive, generally well-tolerated, and low-cost, and findings can be relayed to patients in real-time.^11,12 Studies have confirmed a moderate to strong agreement between ultrasound findings and disease location and extent compared with endoscopy and MRE.^13–15 In adult studies for both CD and UC, the endoscopic findings of active disease have been demonstrated to correlate with IUS bowel wall thickness (BWT) measurements. However, endoscopic scoring systems were not utilized to evaluate the endoscopic disease severity.¹⁶ Furthermore, there is a paucity of pediatric data comparing IUS findings with endoscopic activity in the IBD setting. For that reason, BWT thresholds for determining active inflammation have not been established in children.

Given the possible roles of IUS for assessing the presence of intestinal inflammation at the bedside and monitoring disease activity in pediatric IBD, our primary aim was to evaluate for differences in BWT measured using IUS across different categories of disease severity as documented on ileocolonoscopy. Our secondary aims were to assess the agreement of IUS to colonoscopy for both IBD diagnoses (UC vs CD), assess agreement for disease distribution using Paris Classification,¹⁷ and determine the optimal IUS BWT threshold for detecting bowel wall active inflammation using endoscopy as the referenced standard. We hypothesized that increased BWT is associated with more severe endoscopic inflammatory scoring of the bowel (rated by SES-CD for CD and the UCEIS for UC) in pediatric patients.

Methods

We conducted a single-center cross-sectional pilot study of pediatric patients suspected to have previously undiagnosed IBD, including CD and UC. The study protocol was reviewed and approved by the Children’s Hospital Los Angeles (CHLA) ethics committee, a tertiary pediatric health care center (IRB, CHLA-20-00003). Informed consent and/or assent were obtained from the patient and/or the parent, depending on the subject’s age.

Participants

Patients were recruited from CHLA pediatric outpatient gastroenterology clinics and inpatient units. Consecutive pediatric patients ages 6 to 20 years old suspected of having IBD were prospectively recruited for the study between July 2020 and January 2022. A diagnosis of IBD was suspected based on either a CRP >1 mg/g, a sedimentation rate (ESR) >15 mm/min, unexplained growth failure, and/or a fecal calprotectin >200 ug/g. Exclusion criteria for this study were patients who (1) had a BMI ≥ 99th percentile, (2) refused to proceed with an endoscopic evaluation, or (3) had an endoscopic evaluation more than 4 weeks from the IUS, even if the IUS was performed before the endoscopy.

Forty-nine patients with suspected IBD presenting to the gastroenterology service were offered participation; 13 declined to participate in the study. One patient was excluded, as they had been diagnosed with IBD years prior at an outside institution. Thirty-five patients were enrolled in the study. One patient had to be excluded due to refusing endoscopy after enrollment. A second patient was excluded because of a delay in having an endoscopy beyond 4 weeks after IUS. Thirty-three patients remained in the study for the final analysis. The subject flow diagram is presented in Figure 1.

Figure 1.

Flow diagram of patient participation into the study.

Measures

To describe our study population, we collected demographic information, including age, sex, and baseline symptoms during the enrollment visit, supplemented with chart reviews. We also documented biochemical information, including hemoglobin, hematocrit, albumin, ESR, CRP, and fecal calprotectin level obtained at most ± 3 weeks from the day the IUS was performed. Fecal calprotectin measured within our institution demonstrates an upper threshold of 3000 ug/g, whereas the assays performed outside our institution could measure higher levels. Disease-specific clinical activity scores were collected to assess disease severity at baseline, using the Pediatric Crohn’s Disease Activity Index (PCDAI) for patients with CD and the Pediatric Ulcerative Colitis Activity Index (PUCAI) for patients with UC.

Evaluations of the bowel segments (rectum, sigmoid colon, descending colon, transverse colon, ascending colon, and terminal ileum) were performed using IUS by an experienced and formally IUS-trained gastroenterologist who was blinded to the endoscopic results at the time of the ultrasound examination. No preparation, such as fasting or oral contrast material, was required for the evaluations. All analyses were performed using a Samsung RS85 ultrasound machine (Samsung-Medison; Seoul, Korea) with a low-frequency curvilinear transducer (3-10 MHz) for screening for bowel disease and complications and a high-frequency linear transducer (3-12 MHz) for bowel measurements. To account for changes in BWT related to differences in disease severity and potential bowel wall obliquity, BWT was measured twice in the longitudinal view, 1 cm apart, and twice in the axial (short axis) view. The mean of the 4 measurements was used for analysis, as previously described.¹⁸ Doppler signal in the bowel wall was also collected and reported using a modified Limberg score (grade 0-3, with 0 as no Doppler activity, 1 as a short signal, 2 as a long continuous signal, and 3 as a signal within and outside the bowel wall).¹⁸ The IBD subtype diagnosis (CD vs UC) was determined using IUS data based on the disease distribution (skipped lesions vs continuous inflammation) and the presence of terminal ileal disease. Based on this diagnosis, the Paris Classification for disease distribution was also determined using IUS.

Children also underwent a clinically indicated ileocolonoscopy. The endoscopist, blinded to the ultrasound images, scored the endoscopic severity during the procedure. The IBD subtype diagnosis and the distribution of disease using the Paris Classification was recorded. For patients suspected to have CD on colonoscopy, we collected segmental SES-CD scores. For patients suspected to have UC on colonoscopy, we collected UCEIS scores for each bowel segment.^7,19 Disease distribution was assessed using Paris Classification. Patients with IBD unclassified (IBD-U) were analyzed using the UC Paris Classification and the UCEIS. Both the SES-CD and UCEIS scores were analyzed as categorical variables with 3 levels. An SES-CD score ranged from 0 to 12. An SES-CD score of <3 was considered healthy bowel or inactive CD; a score between 4 and 6 demonstrated mild disease activity; and a score ≥7 indicated moderate to severe activity.²⁰ The UCEIS had a score ranging from 0 to 8. A UCEIS score of 0 or 1 was considered healthy bowel; a score between 2 and 4 was labeled as mild disease activity; and a score of ≥5 demonstrated moderate to severe disease activity.²¹

Statistical Analysis

Statistical analyses were performed using RStudio (version 1.4.1103; RStudio Team, 2020; RStudio, Integrated Development for R. RStudio, PBC, Boston, MA, USA, http://www.rstudio.com/). We evaluated the differences in BWT for bowel segments across disease severities measured by the endoscopic scores and described it as a categorical, ordinal variable with healthy, mild, and moderate to severe disease activity categories. Given the distributions of our data and the small sample size, the nonparametric Kruskal-Wallis test was chosen to assess the differences in the distributions of BWT between endoscopic disease severity categories. For results demonstrating evidence of significant group differences in median BWT, a post hoc analysis was performed using a pairwise Wilcoxon rank-sum test to determine which groups differ from one another, where P values were adjusted for multiple comparisons via the Bonferroni method.

To assess agreement for IBD subtype diagnosis and disease distribution using the Paris Classification between IUS and ileocolonoscopy, we calculated the absolute percent agreement and Cohen’s kappa ($\kappa$) statistic. Unweighted 𝜅 statistics were calculated for nominal variables, such as the IBD subtype. Ordinal variables such as the Paris Classification were analyzed using both unweighted and weighted $\kappa$ statistics, the latter using ordinal weights.

Receiver operating characteristic (ROC) analysis was used to assess the diagnostic performance of IUS BWT for detecting active inflammation on a segmental basis (ie, 6 BWT measurements for each patient) using endoscopy as the reference standard; area under the ROC curve (AUROC), accuracy, sensitivity, and specificity were also calculated. Bowel segments with a segmental SES-CD >2 or a segmental UCEIS >1 were considered actively inflamed for this analysis. The optimal IUS BWT cutoff value was determined using the Youden index²² for all bowel segments, colonic segments independently, and the terminal ileum independently.

Bowel segments with missing BWT measurements were not retained for the primary or secondary analysis. Outlying data points for BWT were validated against corresponding ultrasound images and retained for analysis if they were noted to be an accurate representation of the image. Otherwise, the data point was removed from the analysis. An alpha level of 0.05 was used as the cutoff for statistical significance, and 95% confidence intervals (CIs) were calculated, as appropriate. No formal a priori sample size calculation was performed, as this was a pilot investigation.

Results

Participants Characteristics

A summary of patients and disease characteristics in our sample is provided in Table 1. The cohort’s median age was 15 years old (interquartile range [IQR], 12-16 years), with 12 (36%) females and 21 (64 %) males, and a median body mass index (BMI) z-score of −0.25 (IQR, −1.4 to 0.68). Two patients were 6 years old, and all other patients were 10 years of age or older. The final diagnosis was made by the primary clinical gastroenterologist based on a combination of endoscopy, MRE, and histopathology. Ultimately, 4 patients (12%) had a normal evaluation, despite meeting inclusion criteria. Baseline biochemical profiles included median hemoglobin of 11.8 g/dL (IQR, 10.9-13.0 g/dL), a median CRP of 1.4 mg/dL (IQR, 0.5-3.3 mg/dL), a median ESR of 26 mm/Hr (IQR,16.0-46.5 mm/Hr), and a median albumin 4.0 g/dL (IQR, 3.7-4.4 g/dL). The median fecal calprotectin was 3000 ug/g of feces (IQR, 1409-3000 ug/g of feces). Of note, patients found to have a normal colonoscopy had a median fecal calprotectin of 554 ug/g (IQR, 419.5-720.0 ug/g of feces).

Table 1.

Patient and disease characteristics at inclusion, n = 33.

Patients	n (%), or median (IQR)
Age (year)	15.0 (12.0-16.0)
Female	12 (36.3)
Male	21 (63.6)
Crohn’s disease	16 (48.5)
Ulcerative colitis	11 (33.3)
IBD-unclassified	2 (6.1)
Healthy individuals	4 (12.1)
Colonic disease extent in CD (n = 16)	2 (6.1)
L1	7 (43.8)
L2	3 (18.8)
L3	6 (37.5)
Colonic disease extent in UC/ IBD-U (n = 13)
E1	1 (7.7)
E2	2 (15.3)
E3	3 (23.1)
E4	7 (53.8)
PCDAI at presentation (n = 16)	32.5 (20.0-40.0)
PUCAI at presentation (n = 13)	60.0 (36.3-70.0)
Endoscopic disease activity of bowel segments based on SES-CD (n = 126)
Healthy	97 (77.0)
Mild	18 (14.3)
Moderate/Severe	11 (8.7)
Endoscopic disease activity of bowel segments based on UCEIS (n = 75)
Healthy	23 (30.7)
Mild	40 (53.3)
Moderate/ Severe	12 (16.0)

Abbreviations: L1, Crohn’s of the ileum +/- Cecum; L2, Crohn’s Colitis; L3, Crohn’s Ileocolitis; E2, Ulcerative colitis ad splenic flexure; E3, Ulcerative colitis ad hepatic flexure; E4, Ulcerative pancolitis; PCDAI, Pediatric Crohn’s Disease Activity Index; PUCAI, Pediatric Ulcerative Colitis Activity Index; SES-CD, Simple Endoscopic Score for Crohn’s Disease; UCEIS, Ulcerative Colitis Endoscopic Index of Severity.

Bowel Visualization on Ultrasound

Of the 198 bowel segments, 174 (87.9%) were visible and measurable using IUS. The most difficult segment to assess was the rectum, with 15 of 33 (45.5%) of the patients’ rectums not visualized. Additionally, 1 of 33 (3.0%) of the sigmoid colon, 2 of 33 (6.1%) of the descending colon, 3 of 33 (9.1%) of the transverse colon, and 3 of 33 (9.1%) of the ascending colon segments could not be definitively visualized. The terminal ileum was visualized in all patients. Reasons for poor/non visualization of bowel segments included poor visualization of the rectum through an unfilled bladder, gaseous distention of other bowel segments obscuring portions of the bowel, and obese body habitus.

Prediction of the IBD Diagnosis and Disease Distribution

Based on IUS findings, 16 (48.5%) patients were predicted to have CD, whereas 12 (36.4%) patients were predicted to have UC; and 5 (15.2%) presented with normal findings and therefore were suspected to be healthy or have mild mucosal disease. The absolute percent agreement between IUS prediction of IBD diagnosis and colonoscopy was 69.7%. However, in predicting disease distribution using the Paris Classification, there was only 45.5% agreement between IUS and colonoscopy. Using unweighted $\kappa$ statistics, there was moderate agreement ($\kappa$ = 0.52; 95% CI, 0.28-0.75) between IUS and colonoscopy to determine IBD diagnosis. Using unweighted $\kappa$ statistics, there was a fair agreement ($\kappa$ = 0.36; 95% CI, 0.15-0.56) between modalities for determination of disease distribution using the Paris Classification. Given the ordinal nature of Paris Classifications, we repeated the latter analysis using a weighted version of Cohen’s kappa statistics using ordinal weights, which demonstrated moderate agreement ($\kappa$ = 0.48; 95% CI, 0.14-0.83) between the modalities for determination of disease distribution using the Paris Classification.

Association Between Endoscopic Inflammation Scoring and BWT

The distribution of BWT as measured by IUS across endoscopic severity measured using (1) the SES-CD and (2) the UCEIS is shown in Figures 2A and 2B, respectively.

Figure 2.

Distribution of bowel wall thickness (BWT) as measured by point-of-care intestinal ultrasound, by endoscopic severity measured using either (A) the SES-CD or (B) the UCEIS via ileocolonoscopy using Tukey box plots.

From the segments available for analysis in patients with CD, we found the median BWT for normal (healthy or inactive disease) segments was 1.5 mm (IQR, 1.2-2.0 mm), for mildly inflamed segments 3.2 mm (IQR, 1.7-4.4 mm), and for moderately/severely inflamed segments 5.1 mm (IQR, 4.7-6.6 mm). There was a significant difference in median BWT measurements between the endoscopic severity scores (P < .001). The subsequent pairwise comparison demonstrated a significant difference in measurements between the healthy bowel segments and the moderately/severely inflamed bowel segments (P < .001). No evidence of median difference was observed between healthy bowel segments and mildly inflamed bowel segments (P = .104). The difference in median BWT between mildly inflamed and moderately/severely inflamed bowel segments was borderline significant (P = .06).

In patients with UC, the median BWT for normal (healthy or inactive disease) segments was 1.4 mm (IQR, 1.0-1.6 mm); for mildly inflamed bowel segments it was 2.1 mm (IQR, 1.6-2.8 mm); and for moderately/severely inflamed bowel segments it was 3.3 mm (IQR, 3.1-3.7 mm). There was a significant difference in median BWT measurements between the endoscopic severity scores in patients with UC (P < .01). For this disease process, pairwise comparison demonstrated a significant difference in measurements between the healthy bowel segments and the moderately/severely inflamed segments (P = .001) and between the mildly inflamed bowel segments and moderately/severely inflamed bowel segments (P = .006). There was no evidence of median difference between the BWT of healthy bowel segments and mildly active bowel wall segments (P = .17) in patients with UC.

Determination of BWT Threshold Predictive of Endoscopic Inflammation

Figure 3 demonstrates the ROC curve computed from all BWT measurements in all bowel wall segments combined (AUROC, 0.743; 95% CI, 0.67-0.82). We determined that an optimal cutoff BWT measuring 1.9 mm offered the best trade-off between sensitivity (64%; 95% CI, 53%-73%) and specificity (76%; 95% CI, 65%-85%) in detecting an actively inflamed bowel segment on colonoscopy. When assessing colonic segments alone, we obtained an AUROC of 0.706 (95% CI, 0.62-0.79). The optimal BWT measurement cutoff for colonic segments remained 1.9 mm, with a sensitivity of 58% (95% CI, 47%-69%) and a specificity of 74% (95% CI, 62%-84%). However, when evaluating only the terminal ileum, we obtained an AUROC of 0.88 (95% CI, 0.74-1.00). The terminal ileum had a similar optimal BWT cutoff value of 1.9 mm, providing a higher sensitivity of 93% (95% CI, 68%-100%) and a higher specificity of 83% (95% CI, 59%-96%) for detecting active intestinal inflammation.

Figure 3.

Receiver operator characteristic (ROC) curve for intestinal ultrasound (IUS) as a predictor of disease activity seen on individual bowel wall segments as compared with disease activity on ileocolonoscopy for (A) all bowel segments, (B) isolating colonic segments, and (C) isolating terminal ileum segments.

Discussion

Intestinal ultrasound is a noninvasive imaging tool that can assess bowel inflammation during a gastroenterology clinic visit. Our study demonstrates that IUS performed at the bedside was effective in visualizing over 85% of bowel wall segments, with the rectum being the most difficult to see using transabdominal methods. Our study also demonstrated that IUS can suggest a diagnosis of CD vs UC based on disease distribution. However, comparing the prediction of the exact disease extent on IUS using the Paris Classification compared with endoscopy findings only offered a fair to moderate agreement. Relative to standardized endoscopic scores, bowel segments with more severe inflammation had significantly greater BWT measurements on IUS for patients with either UC or CD. Intestinal ultrasound provided an AUROC of 0.74 to discriminate between the presence or absence of inflammation on ileocolonoscopy. The optimal cutoff BWT value when maximizing both sensitivity and specificity for separating healthy bowel from pediatric IBD was 1.9 mm for all bowel segments.

Our study assessed the ability to predict CD vs UC diagnosis as determined by ileocolonoscopy. When we used disease distribution on IUS (such as involvement of the terminal ileum or skip lesions), IUS accurately identified the IBD phenotype (CD vs UC) in 70% of the patients, which is modest. It also demonstrated a moderate agreement compared with ileocolonoscopy. However, there was both lower accuracy and agreement between IUS and ileocolonoscopy for the determination of the Paris Classification. Although multiple studies have demonstrated a high agreement between IUS and colonoscopy for the presence of inflammation, few studies describe the agreement between the 2 modalities for disease extent.²³ Therefore, in undifferentiated patients with a suspicion of IBD, our study signals that IUS could be used to predict the presence of inflammation and the disease phenotype. In addition, others have demonstrated that IUS can help determine disease distribution in the setting of an incomplete colonoscopy.²⁴ Currently, there is insufficient evidence in children to predict whether early IUS could provide a baseline for assessing the timely response to IBD medical therapy via repeated measurement of BWT. Our study demonstrates that IUS effectively visualizes most bowel segments in children with suspected IBD. However, in our pilot study, this approach is less effective at observing the rectum, suggesting there may be limited utility for tracking patients with isolated proctitis. Similar results were noted in adult studies, where the rectum was generally poorly visualized.²⁵ Using a transperineal and/or transrectal ultrasound approach is a more optimal technique in the assessment of the rectum and perineal area.²⁶

In pediatric patients with UC, increased BWT has been associated with increased severity of endoscopic appearance using the Mayo endoscopic score.²⁷ Our study similarly demonstrated an association between BWT and validated endoscopic scores for patients with UC and CD from IUS exams performed in the gastroenterology clinic. Indeed, increasing BWT in our sample was associated with increasing disease severity on colonoscopy. The value of IUS may be even greater in the setting of severe disease, where patients may have incomplete ileocolonoscopies. This noninvasive tool can evaluate disease severity and distribution in a timely manner without the need for contrast.²⁴ However, similar to other studies, discrimination of BWT was best in differentiating between healthy bowel segments and moderate to severe disease.^27,28 This indicates that IUS may appear normal in patients with mild disease activity or superficial disease, which may not lead to significant transmural damage. On the other hand, some segments demonstrate transmural inflammation without significant mucosal disease. It leads us to question the underlying mechanism of transmural disease compared with mucosal disease. Previous research has demonstrated that inflammation can be confined to the intramural portion of the bowel and not be evident on endoscopy (ie, the mucosa appears normal or healed), particularly in CD.^29,30 In the current study, this phenomenon may account for the wide distribution of BWT measurements seen in the healthy-appearing bowel segments in patients with CD compared with the distribution of BWT measurements in healthy segments of patients with UC.

Currently, the BWT indicative of active inflammation accepted for both pediatric and adult patients is 3 mm. However, BWT thresholds have not been specifically established in pediatric patients. Studies have demonstrated that healthy children under age 15 have BWT on IUS under 2 mm; this slightly increases with age.^31–33 We demonstrated through this study that in children, using a BWT >2 mm may be more appropriate in the pediatric population, offering the best trade-off between sensitivity of 64% and a specificity of 76%. Notably, this degree of BWT would assist in detecting patients with mild disease. In addition, we had an excellent performance in evaluating disease of the terminal ileum, with using a 2 mm cutoff, offering a sensitivity of 93% and a specificity of 83%. This implies that ultrasound may be a perfect tool to detect ileal disease. There is a possibility that older age groups demonstrate a decrease in specificity using this BWT cutoff value. Validation within a larger cohort with varying age groups and containing adequate numbers of patients with both CD and UC is needed to definitively establish the ideal BWT upper limits of normalcy in distinguishing healthy bowel from actively inflamed bowel.

This study has several limitations. First, the ultrasound assessment was performed by a single observer, which brings to question the generalizability of the results. However, the sonographer, a fellowship-trained pediatric gastroenterologist, applied a systematic approach to IUS scanning using internationally accepted methodology to increase reproducibility. Additionally, multiple studies have demonstrated substantial to almost perfect interobserver agreement in BWT measurements among gastroenterologists using it in a point-of-care setting^.13,34 Second, this study has a relatively small sample size, and therefore the results were presented as an evaluation of bowel segments as units rather than patients as units. Using bowel segments as the subject units is also another way to address a biased patient population, as the patients included had a high pretest likelihood of having disease activity. Although the patients were undifferentiated at the time of the diagnostic testing, they had a high probability of having active disease. Nevertheless, our analysis highlights that a higher BWT seen on IUS is associated with increasing severity of inflammation on endoscopy. Third, since there was a low incidence of stricturing disease, internal penetrating disease, and abscess formation, we were unable to address the performance of IUS in detecting disease complications. And finally, young patients under the age of 10 years old were underrepresented in the cohort, and therefore age-specific findings could not be addressed. Therefore, the performance of IUS still needs to be investigated in young age groups, particularly in patients with very early onset IBD.

Conclusion

In pediatric patients with IBD, we found that IUS had a moderate to good agreement in differentiating an IBD diagnosis of CD vs UC, that the degree of BWT is associated with disease severity, and that pediatric patients may benefit from lower cutoffs to define presence of inflammation of a bowel segment. Evaluation of the terminal ileum had particularly high sensitivity and specificity for detecting the presence of disease. These findings indicate there may be a use for point-of-care IUS as an evaluation tool of disease activity and severity in a North American pediatric clinical practice. Further studies are needed to validate these findings in larger pediatric IBD cohorts.

Glossary

Abbreviations

AUROC

area under the receiver operator curve

CD

Crohn’s disease

BWT

bowel wall thickness

CRP

C-reactive protein

ESR

sedimentation rate

$\kappa$

Cohen’s kappa;

IBD

inflammatory bowel diseases

IQR

interquartile range

IUS

intestinal ultrasound

MRI

magnetic resonance imaging

SES-CD

Simple Endoscopic Score for Crohn’s disease

UC

ulcerative colitis

UCEIS

Ulcerative Colitis Endoscopic Index of Severity

Author Contributions

M.C.: First author and primary writer conceptualized and designed the study, completed patient recruitment, and performed the data analysis.

L.H.: The second author and primary writer conceptualized the study.

P.H.R.: Data analysis, manuscript review, and primary editor.

J.R.D.: Expert in the field of intestinal ultrasound and abdominal imaging. Conceptualized and designed the study, reviewed data, and edited the final manuscript.

D.B.P.: Expert in the field of inflammatory bowel disease. Conceptualized and designed the study, reviewed data, and edited the final manuscript.

Funding

This work was supported by the National Center for Advancing Translational Sciences (NCATS) Mentored Career Development in Clinical and Translational Research Award (KL2; Grant number: KL2TR001854) and by the American Gastroenterology Association Pfizer Pilot Research Award in Inflammatory Bowel Disease (grant number: AGA2020-21-03).

Conflicts of Interest

J.R.D. has unrelated ultrasound research support from Siemens Medical Solutions USA, Canon Medical Systems, and Bracco Diagnostics. All other authors have no conflict of interest to declare.