Crohn disease (CD) is a transmural disease that can affect any location in the intestinal tract. Evaluation of the small bowel is essential in CD because upper endoscopy and colonoscopy evaluate only a small portion of the small intestine (SI). Therefore, clinicians rely on diagnostic imaging and video capsule endoscopy to evaluate the presence and extent of disease in the SI.
Traditional small intestinal imaging includes small bowel follow-through (SBFT) and computed tomography (CT) scans, both exposing patients to ionizing radiation (1). Ionizing radiation resulting from medical imaging has increased dramatically during the last decade, and even low radiation doses in children increase their risk for radiation-induced cancer (2). CD is associated with high exposure to ionizing radiation resulting from repeated medical imaging (3,4).
Magnetic resonance enterography (MRE) is an imaging technique that is free of radiation exposure and yields excellent images of the small and large intestine. MRE is able to differentiate active inflammation from chronic inflammation in the layers of the bowel wall; however, there are few data comparing MRE with endoscopy, histopathology, and laboratory data, which are objective measures of disease.
The purpose of the present study is to compare MRE findings with objective measures of CD, including laboratory data and endoscopy/histopathology. We did not seek to determine a sensitivity and specificity for diagnosis, which has been done (5), nor did we seek to compare MRE with subjective clinical symptoms or clinical disease activity measures.
An institutional review board–approved query of our CD MRE database maintained prospectively since 2008 was performed. Patient demographics, diagnosis, date of diagnosis, laboratory evaluation, endoscopy findings, and histopathology results were collected.
Children underwent MRE studies at the discretion of their primary pediatric gastroenterologist for symptomatic disease either at diagnosis or during an acute flare. Most children underwent MRE for abdominal pain/diarrhea. Some children underwent MRE at diagnosis to evaluate small bowel disease in lieu of other SI imaging modalities such as an SBFT. Patients undergoing pelvis MRI only to evaluate perianal disease were not included in the present study. All of the children underwent MRE evaluation at Children's Healthcare of Atlanta.
MRE PROTOCOL AND RESULTS
All of the MRE scans were performed on a present-generation 3 Tesla MRI system (Magnetom Trio, Siemens Medical Solutions, Iselin, NJ). Preprocedure patient preparation was used in all of our outpatient, nonsedated patients. One bottle (450 cm3) of VoLumen (0.1% barium sulfate with sorbitol) oral contrast was administered in these patients 90 minutes before imaging and a second bottle 30 minutes before imaging. For younger patients requiring sedation, patients drank Omnipaque mixed in sugar-free Kool-Aid using an age-based chart, which was the standard protocol for our CT scans using oral contrast.
Our bowel MRI protocol consists of a combination of T2-weighted (T2W) and dynamic, multiphase contrast-enhanced T1-weighted (T1W) images through the abdomen and pelvis, extending from the lung bases through the perineum. T2W images were acquired with a single-shot fast-spin echo (ssT2) sequence (380 mm2 field of view (FOV), 320 × 224 matrix, 5-mm slice thickness, repetition time/echo time (TR/TE)/flip angle = 1500/70/150 degree, acceleration factor of 2.0) in the coronal and axial plane without fat saturation, and in the axial plane with fat saturation using spectral adiabatic inversion recovery technique (SPAIR). An axial steady-state free precession sequence was also obtained with a single-shot technique (380 mm2 FOV, 320 × 256 matrix, TR/TE/flip angle = 3.51/1.55/50 degree, 50 slices at 5-mm thickness (no gap) and bandwidth of 1040 Hz/pixel). A coronal, thick section heavily T2W sequence (380 mm2 FOV, 384 × 307 matrix, 60-mm slice thickness, TR/TE/flip angle = 5000/709/180 degree, acceleration factor of 2.0) is then acquired 15 times (each slab acquisition between 2 and 5 seconds) to assess the dynamic properties of bowel motility. Additional T2W sequences were also acquired through the pelvis with ssT2 sequence in the sagittal plane and a high-resolution, 3-dimensional (3D) turbo spin echo sequence using variable flip angles in the axial plane (250 mm2 FOV, 256 × 257 matrix, 1-mm slice thickness, TR/TE/flip angle = 1300/97/variable degree, approximately 5–6 minutes acquisition time).
3D T1W gradient echo images were obtained in the precontrast phase (axial and coronal), and then subsequently in the arterial, venous, and (axial and coronal) delayed phases with the following parameters: 380 mm2 FOV, 288 matrix (80% phase resolution), partial Fourier imaging 6/8, TR/TE/flip angle = 3.44/1.25/10 degree, 104 slices at 2.8-mm slice thickness and bandwidth at 450 Hz/pixel and 2× acceleration. Each patient's weight was recorded and gadopentetate dimeglumine (Magnevist, Bayer Schering Pharma, Montville, NJ) was administered at a dose of 0.1 mmol/kg and a rate of 2 cm3/second, followed by a 20-cm3 saline flush at 2 cm3/second using a dual-chamber power injector (Spectris, Medrad, Warrendale, PA). Acquisition time for 3D gradient echo images was 15 to 17 seconds, performed in a single breath hold. Our protocol does not use antimotility agents and does not include enteroclysis aimed at bowel distention.
More important, our protocol does not include enteroclysis or antimotility agents, yet yields excellent images. No children needed a repeat study for poor images in our population. The results of each MRE were confirmed by a pediatric radiologist with specific experience and additional training in body MRI (U.U. or A.A.) who was not provided with laboratory or histology results. Laboratory results were typically reached in the outpatient laboratory and were not provided nor are they available in the electronic medical record. MRE results were separated into 3 categories, including active inflammation, chronic disease changes without inflammation, and normal bowel. These categories are defined below as described in the literature (6–8). Some children demonstrated both active inflammation and chronic disease on different bowel segments.
MRE active inflammation was defined as bowel wall thickening and enhancement on postgadolinium T1W images PLUS high signal intensity on T2W-SPAIR fat-suppressed images. MRE chronic disease changes were defined as bowel wall thickening and enhancement on postgadolinium T1W images PLUS low signal intensity on T2W-SPAIR fat-suppressed images. Normal bowel was defined as absence of bowel wall thickening and enhancement on postgadolinium T1W.
We chose these categories because active inflammation often requires treatment changes, whereas chronic disease changes likely represent remodeling of active inflammation, may represent fibrosis, and, most important, are without active inflammation. Children could be classified in both active and chronic groups if different bowel segments displayed these findings. It should also be noted that active inflammation can mask chronic disease changes, which can only be differentiated after treatment. We chose not to include lymph node enlargement, comb sign, free fluid, or other less objective measures of inflammation because these may not represent disease attributable to CD. Furthermore, the significance of these isolated findings without intestinal inflammation is unknown.
Inclusion and Exclusion Criteria
All of the children diagnosed as having CD who underwent MRE were included in the study with regard to MRE results. Not all of the children included underwent laboratory evaluation within 14 days or endoscopy within 30 days and therefore were not evaluated in these comparisons. Children diagnosed as having ulcerative colitis were not included in the present study.
Histology and Confirmation of CD
All of the children in the study were diagnosed as having CD based on clinical presentation and histology by their primary gastroenterologist according to national guidelines (9). Biopsies were reviewed by a pathologist and classified as either normal or inflamed terminal ileum (TI) biopsies and normal or inflamed colonic biopsies for comparison with MRE. Upper endoscopy (duodenum) biopsies were also reviewed as part of diagnosis but were not included in this evaluation of MRE.
Comparison of MRE With Laboratory Evaluation
All of the patients who had laboratory evaluation within 14 days of MRE underwent comparison between MRE findings and laboratory results. Patients were separated into 2 categories by MRE: presence or absence of active inflammation. Albumin, C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), and hemoglobin were compared between groups.
Comparison of MRE With Endoscopy/Histology Evaluation
All of the patients who had endoscopy and histology within 30 days of MRE underwent comparison between MRE findings and endoscopy/histopathology results. Endoscopy results were classified as ulcers, erythema/friability without ulcers, and normal for both the TI and colon with the most involved portion being classified. Histology results of the TI and colon were separated into active inflammation or normal. Active inflammation was reported by pathologists and was defined by an increase in inflammatory content within the lamina propria of biopsy specimens. MRE findings were described by 2 segments (TI or colon) and labeled as having either presence or absence of active inflammation. Comparisons were made between MRE findings and histology results.
Statistical analyses were performed using SAS 9.2 (SAS Institute, Cary, NC). Statistical significance was assessed using a significance level of 0.05. Proportions were computed to summarize categorical data, whereas means and medians were calculated for continuous measures. When distributional assumption was not valid, medians along with the range were calculated. Subjects were categorized into 2 groups based on their MRE findings: MRE active inflammation in the SI and/or colon and absence of active inflammation (normal bowel). Comparisons between each of these categories and laboratory results (albumin, hemoglobin, ESR, CRP) were made (Mann-Whitney test) and P values reported. To measure agreement between MRE and endoscopy, the κ statistic and associated 95% confidence intervals were computed. Because of the small cell sizes, exact P values are reported. In the present study, the following guidelines proposed by Landis and Koch in 1977 are used to interpret measures of agreement: 0.01 to 0.20, slight agreement; 0.21 to 0.40, fair agreement; 0.41 to 0.60, moderate agreement; 0.61 to 0.80, substantial agreement; and 0.81 to 1.00, almost perfect agreement.
A total of 147 MRE studies were performed in 119 children with CD (median 1, range 1–4). The mean age of diagnosis for CD was 12.2 years (SD 3.7 years), and the mean disease duration at time of MRE was 1.58 years (SD 2.0 years). The population was 51% white (n = 59), 44% African American (n = 51), and 4% other (n = 5). The number of boys (n = 65) was higher than the number of girls (n = 54) (Table 1).
Total imaging time was 25 to 60 minutes, depending on cooperation of child and the need for sedation. Seventeen MRE studies required sedation, which was performed after ingestion of contrast. The youngest child was 3 years old, and sedation was used for children younger than 8 years. Sedation was performed by a pediatric anesthesiologist and the method used was at their discretion, but typically included general anesthesia. The examination could be stopped at any time for patient discomfort, difficulty with being enclosed in an MR scanner, or other symptoms, although in our cohort the examination was not stopped in any patient.
SI MRE findings included 24 studies demonstrating active inflammation without chronic disease changes (Fig. 1), 47 studies displaying both active and chronic changes in different segments, and 12 studies with only chronic disease changes (Fig. 2), whereas 64 studies demonstrated a normal SI. Meanwhile, colon MRE findings included 35 studies displaying active inflammation without chronic disease, 28 studies with active and chronic disease, 6 studies demonstrating chronic disease only, and 78 studies with a normal colon. On a per-patient basis, a total of 30 (20.4%) studies had a normal SI and colon, 47 (32.0%) studies displayed disease in the SI only, 33 (22.4%) studies displayed disease in the colon only, and 37 (25.2%) studies demonstrated disease in both the SI and colon (Table 2).
A total of 32 children underwent MRE within 1 week of diagnosis, 43 total children within 1 month, and 53 total children within 3 months of diagnosis for evaluation of small bowel disease. Treatment was started in all of the children at the discretion of their primary gastroenterologist. Of the 32 children undergoing MRE within 1 week, 30 were not started on treatment before MRE. Of the 21 children who underwent MRE within 3 months of diagnosis but not within 1 week, 18 children were begun on treatment before MRE. Of the 53 total children, 47 children displayed active inflammation in either the SI or colon. The 6 children with a normal MRE had MRE studies within 1 month and all 6 displayed only mild mucosal disease on endoscopy without ulcers. One diagnosis was based on granulomatous esophageal lesions, another diagnosis was based on pyoderma gangrenosum, and the other 4 were diagnosed as having CD based on erythema/friability on endoscopy with typical CD pathology on biopsy. Interestingly, 4 of the 6 children were younger than 10 years.
Comparison of MRE With Laboratory Evaluation
Laboratory data were available in 88 children within 14 days of MRE. We compared laboratory values of children with active inflammation on MRE (n = 69) with children with no active inflammation on MRE (n = 19). There was a statistically significant difference in CRP with a mean CRP of 3.6 in the MRE active inflammation group and a mean CRP of 1.1 in the MRE normal bowel group (P < 0.001). There was also a higher ESR in the active inflammation group (36 vs 22, P = 0.031), higher platelets (439 vs 352, P = 0.033), and lower albumin (3.4 vs 3.7, P = 0.049), but no difference in hemoglobin (Table 3).
Comparison of MRE With Endoscopy and Histology
Endoscopy was performed within 30 days of MRE in 67 children with 49 ileal biopsies and 67 colonic biopsies. The most common reasons for failure to intubate the ileum were severe colonic disease (n = 15) and technical difficulty (n = 3). Although each colonoscopy included multiple colon biopsies, we evaluated only the most diseases segment of the colon based on endoscopic disease.
Ileal endoscopy demonstrated 20 children with ulcers, all with active inflammation on MRE; 20 children with normal TI, 19 of which had no active inflammation on MRE; and 9 children with only erythema and/or friability on endoscopy. Three displayed active inflammation on MRE and 6 had a normal MRE. There was excellent agreement between endoscopy and MRE when ulcers were present (κ = 0.95; P < 0.001), and poor agreement when only erythema or friability was present (κ = 0.334; P = 0.076) (Table 4).
In 24 children with active ileal inflammation on MRE, 21 displayed inflammation on histopathology, whereas 3 had normal ileal histology. In 25 children with no active inflammation on MRE, 21 children displayed normal histology, whereas 4 demonstrated inflammation on ileal biopsies. There was moderate agreement between MRE and histology of the TI with a κ of 0.71 (P ≤ 0.001) (Table 4).
Colonoscopy demonstrated ulcers in 35 children with 34 demonstrating active inflammation on MRE, whereas 1 child had a normal MRE without active inflammation. Endoscopy demonstrated normal findings in 16 children, all with a normal MRE of the colon. A total of 16 children demonstrated erythema/friability on colonoscopy, and 15 of those children had a normal MRE. There was excellent agreement between active inflammation on MRE and ulcers on colonoscopy (κ = 0.955, P < 0.001) but poor agreement between MRE and colonoscopy when only erythema and friability (mild mucosal disease) were present (κ = 0.063, P = 1.0) (Table 5).
Histopathology demonstrated colonic inflammation in 51 patients and 35 of those who displayed active inflammation of their colon on MRE, whereas 16 demonstrated inflammation on histopathology with a normal MRE. Of these 16 patients, 15 had mild mucosal disease (erythema/friability) on colonoscopy. Colon histology was normal in 16 children, all with normal MRE studies. There was fair agreement between MRE and histology of the colon with a κ of 0.511 (P < 0.001) (Table 5).
We report a significant association between MRE findings of active inflammation and CRP, ESR, platelets, and albumin in children with CD, but no association with hemoglobin. Children with active inflammation on MRE displayed a higher CRP, ESR, and platelets, all of which are expected to be elevated in inflammatory states. Active inflammation on MRE was also associated with lower albumin. The present study supports the literature demonstrating a correlation between MRE inflammation and CRP (10–12), but it adds additional data to suggest that other typical laboratory abnormalities in CD (low albumin, high platelet count, high ESR) are associated with active inflammation on MRE. Because of the subjective nature of clinical activity measures and clinical symptoms, we did not compare MRE findings to these measures because it has been done in other studies with conflicting results (10,11,13–16). Although association with laboratory findings may suggest that MRE does not provide useful additional information, we suggest that the association confirms that MRE measures disease activity, and that MRE is able to objectively document disease location, evaluate for complications, and provide a basis for monitoring improvement of disease activity or complications if necessary, which adds significant important medical information for the management of CD.
We also report excellent agreement between endoscopy of the TI and colon with MRE inflammation when ulcers were present. There was also moderate agreement with MRE and histopathology, with most disagreement relating to mild mucosal disease with erythema/friability without ulcers (Fig. 3). These findings demonstrate the ability of MRE to detect active inflammation, defined by endoscopy and histology. The excellent agreement with MRE inflammation with the presence of ulcers on endoscopy is important because it can be extrapolated to small bowel lesions, for which there is no criterion standard for evaluation of disease burden in CD. These data add to the literature suggesting that MRE can detect active inflammation as identified by pathology/histology (6,17) and endoscopy (18). More important, this confirms the ability of MRE to detect moderate to severe inflammation described by ulcers, and confirms the inability of MRE to detect only mild mucosal inflammation described by erythema or friability but without ulceration. These data highlight the ability of MRE to detect significant mucosal disease described by endoscopy, which suggests the use of MRE to detect significant mucosal disease in the SI, not accessible by standard endoscopy, and suggests the use of MRE to objectively evaluate small bowel disease.
All of the patients who underwent an MRE were symptomatic; however, 30 children demonstrated completely normal MRE findings in the small bowel and colon, all of which were performed at follow-up (32%). Almost one-third of our population undergoing MRE for typical flare symptoms did not have objective documentation of active inflammation by MRE. Symptoms, especially in children, may not necessarily represent true disease activity. All of the MRE studies were performed for clinical reasons, given a high clinical suspicion for intestinal inflammation. Thus, one would expect a high number of abnormal studies, yet 32% of our follow-up studies demonstrated normal findings in the bowel. A normal MRE remains an important piece of medical information confirming the absence of active inflammation and disease complications. We suggest that when symptomatic at follow-up, children should undergo objective confirmation of disease before treatment changes are initiated.
We report MRE findings of active inflammation of the small bowel in 47 of 53 newly diagnosed patients with CD, whereas all of the diagnosis MRE studies demonstrated active inflammation in either the colon or SI. Diagnostic MRE studies were used to assess the degree of small bowel involvement in newly diagnosed CD. These data build on recent literature suggesting that MRE has good sensitivity and specificity for diagnosing CD in children (5). Other imaging modalities to evaluate the extent of small bowel involvement in CD, such as small bowel series or abdominal CT imaging, expose children to harmful radiation. Although wireless capsule endoscopy (WCE) displays a higher sensitivity at detecting mild mucosal disease (19,20), it lacks the ability to detect major complications and may be difficult to monitor improvement because location is often uncertain. We suggest that the ability to detect major complications may overcome a slight decreased sensitivity to mucosal-only disease, and that ultimately WCE and MRE may be complimentary studies used to evaluate SI mucosal disease (WCE), transmural inflammation (MRE), and unknown complications (MRE) in CD.
The study is limited by its retrospective design, which undoubtedly introduces selection bias. Our institution-wide best practices guidelines that suggest MRE imaging at diagnosis and objective disease evaluation with endoscopy and/or MRE with acute flares depending on disease location at diagnosis limit, but do not eliminate, this selection bias. At our institution, >90% of children with newly diagnosed CD undergo MRE evaluation of the SI instead of SBFT or other imaging. MRE was confirmed by a radiologist (A.A./U.U.) with an interest and training in body MRE who was blinded to clinical data in an attempt to limit confounding factors of clinical data known by the initial radiologist. Finally, not all of the children who underwent an MRE had other disease activity measurements, including laboratory examination and endoscopy, although children with abnormal laboratory evaluation were more likely to undergo endoscopy, which may introduce additional bias. Future studies with comprehensive disease evaluation including laboratory evaluation, endoscopy, and MRE may provide additional information.
Our study demonstrates that active inflammation on MRE correlates with other objective markers of inflammation, including CRP, ESR, and platelets. Furthermore, we demonstrate excellent agreement between MRE and endoscopy when ulcers are present, suggesting that MRE can detect moderate to severe mucosal disease and that bowel wall edema and inflammation is common in children with ulcers, as would be expected in CD, which is characteristically transmural. Furthermore, our data suggest the use of MRE to objectively document small bowel and colon inflammation in CD, thereby potentially avoiding the use of colonoscopy, especially if small bowel disease is prominent, which would not be accessible by colonoscopy. These findings add to a growing body of work demonstrating the ability of MRE to detect active inflammation and may further suggest a role for MRE in the monitoring of disease activity in CD (5,6,17,18,21). We suggest that a normal MRE in the face of symptomatic disease is important medical information. Research to better define the role of MRE in children with CD is necessary.
1. Gaca AM, Jaffe TA, Delaney S, et al. Radiation doses from small-bowel follow-through and abdomen/pelvis MDCT in pediatric Crohn disease. Pediatr Radiol
2. Brenner DJ, Hall EJ. Computed tomography—an increasing source of radiation exposure. N Engl J Med
3. Desmond AN, O’Regan K, Curran C, et al. Crohn's disease: factors associated with exposure to high levels of diagnostic radiation. Gut
4. Sauer CG, Kugathasan S, Martin DR, et al. Medical radiation exposure in children with inflammatory bowel disease estimates high cumulative doses. Inflamm Bowel Dis
5. Horsthuis K, de Ridder L, Smets AM, et al. Magnetic resonance enterography for suspected inflammatory bowel disease in a pediatric population. J Pediatr Gastroenterol Nutr
6. Lawrance IC, Welman CJ, Shipman P, et al. Correlation of MRI-determined small bowel Crohn's disease categories with medical response and surgical pathology. World J Gastroenterol
7. Parisinos CA, McIntyre VE, Heron T, et al. Magnetic resonance follow-through imaging for evaluation of disease activity in ileal Crohn's disease: an observational, retrospective cohort study. Inflamm Bowel Dis
8. Zappa M, Stefanescu C, Cazals-Hatem D, et al. Which magnetic resonance imaging findings accurately evaluate inflammation in small bowel Crohn's disease? A retrospective comparison with surgical pathologic analysis. Inflamm Bowel Dis
9. Bousvaros A, Antonioli DA, Colletti RB, et al. Differentiating ulcerative colitis from Crohn disease in children and young adults: report of a working group of the North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition and the Crohn's and Colitis Foundation of America. J Pediatr Gastroenterol Nutr
10. Alexopoulou E, Roma E, Loggitsi D, et al. Magnetic resonance imaging of the small bowel in children with idiopathic inflammatory bowel disease: evaluation of disease activity. Pediatr Radiol
11. Del Vescovo R, Sansoni I, Caviglia R, et al. Dynamic contrast enhanced magnetic resonance imaging of the terminal ileum: differentiation of activity of Crohn's disease. Abdom Imaging
12. Maccioni F, Viscido A, Broglia L, et al. Evaluation of Crohn disease activity with magnetic resonance imaging. Abdom Imaging
13. Laghi A, Borrelli O, Paolantonio P, et al. Contrast enhanced magnetic resonance imaging of the terminal ileum in children with Crohn's disease. Gut
14. Florie J, Wasser MN, Arts-Cieslik K, et al. Dynamic contrast-enhanced MRI of the bowel wall for assessment of disease activity in Crohn's disease. AJR Am J Roentgenol
15. Koh DM, Miao Y, Chinn RJ, et al. MR imaging evaluation of the activity of Crohn's disease. AJR Am J Roentgenol
16. Schunk K, Kern A, Oberholzer K, et al. Hydro-MRI in Crohn's disease: appraisal of disease activity. Invest Radiol
17. Punwani S, Rodriguez-Justo M, Bainbridge A, et al. Mural inflammation in Crohn disease: location-matched histologic validation of MR imaging features. Radiology
18. Dagia C, Ditchfield M, Kean M, et al. Feasibility of 3-T MRI for the evaluation of Crohn disease in children. Pediatr Radiol
19. Jensen MD, Nathan T, Rafaelsen SR, et al. Diagnostic accuracy of capsule endoscopy for small bowel Crohn's disease is superior to that of MR enterography or CT enterography. Clin Gastroenterol Hepatol
20. Tillack C, Seiderer J, Brand S, et al. Correlation of magnetic resonance enteroclysis (MRE) and wireless capsule endoscopy (CE) in the diagnosis of small bowel lesions in Crohn's disease. Inflamm Bowel Dis
21. Panes J, Bouzas R, Chaparro M, et al. Systematic review: the use of ultrasonography, computed tomography and magnetic resonance imaging for the diagnosis, assessment of activity and abdominal complications of Crohn's disease. Aliment Pharmacol Ther