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

Authors


  • This uncommissioned systematic review was subject to full peer-review.

Dr J. Panés, Department of Gastroenterology, Hospital Clínic of Barcelona, Villarroel 170, 08036 Barcelona, Spain.
E-mail: jpanes@clinic.ub.es

Abstract

Aliment Pharmacol Ther 2011; 34: 125–145

Summary

Background  Cross-sectional imaging techniques, including ultrasonography (US), computed tomography (CT) and magnetic resonance imaging (MRI), are increasingly used for evaluation of Crohn’s disease (CD).

Aim  To perform an assessment of the diagnostic accuracy of cross-sectional imaging techniques for diagnosis of CD, evaluation of disease extension and activity and diagnosis of complications, and to provide recommendations for their optimal use.

Methods  Relevant publications were identified by literature search and selected based on predefined quality parameters, including a prospective design, sample size and reference standard. A total of 68 publications were chosen.

Results  Ultrasonography is an accurate technique for diagnosis of suspected CD and for evaluation of disease activity (sensitivity 0.84, specificity 0.92), is widely available and non-invasive, but its accuracy is lower for disease proximal to the terminal ileum. MRI has a high diagnostic accuracy for the diagnosis of suspected CD and for evaluation of disease extension and activity (sensitivity 0.93, specificity 0.90), and is less dependent on the examiner and disease location compared with US. CT has a similar accuracy to MRI for assessment of disease extension and activity. The three techniques have a high accuracy for identification of fistulas, abscesses and stenosis (sensitivities and specificities >0.80), although US has false positive results for abscesses. As a result of the lack of radiation, US or MRI should be preferred over CT, particularly in young patients.

Conclusions  Cross-sectional imaging techniques have a high accuracy for evaluation of suspected and established CD, reliably measure disease severity and complications; they may offer the possibility to monitor disease progression.

Introduction

Assessment of intestinal inflammatory lesions by imaging techniques is essential for management of patients with Crohn’s disease (CD). Characterisation of disease location and extension at the time of diagnosis is required to establish a proper management plan. Imaging techniques are also the accepted reference for detection of complications including strictures and penetrating lesions such as fistulas and abscesses. Recently, awareness of the shortcomings of mere clinical evaluation for assessment of disease activity has grown. Some patients with an established diagnosis of CD and symptoms compatible with a disease flare do not have evidence of active CD by laboratory, endoscopic and radiologic criteria. In two large clinical trials, 18% of patients with CD and moderate-to-severe clinical symptoms had no evidence of ulcer at ileocolonoscopy.1, 2 On the other hand, a sizable proportion of patients with established CD may have persistent severe lesions in the absence of symptoms3 and persistence of lesions is associated with higher requirements for hospitalisation and surgeries.4 Therefore, objective assessment of inflammatory lesions is required for guiding therapeutic interventions and for assessing the efficacy of these interventions.

Ileocolonoscopy has been the gold standard for evaluation of lesions in the colon and terminal ileum. However, ileocolonoscopy cannot always be complete and there are several drawbacks related to the invasiveness, procedure-related discomfort, risk of bowel perforation and relatively poor patient acceptance. Over the last few years, cross-sectional imaging techniques, including ultrasonography (US), computed tomography (CT) and magnetic resonance imaging (MRI), have been increasingly used for evaluation of patients with CD. The choice between imaging techniques is often determined by local availability and expertise and technical details of these examinations are also subject to considerable variation, which may affect accuracy.

The purpose of the present review was to provide an objective assessment of the diagnostic accuracy of cross-sectional imaging techniques in three different scenarios: (i) for diagnosis of CD; (ii) for assessment of disease extension and activity; and (iii) for diagnosis of complications. Recommendations on the use of imaging techniques provided in this review are based on an analysis of published evidence selected according to quality criteria, performed by six gastroenterologists with experience in inflammatory bowel disease (IBD) and five radiologists with experience in the use of US, CT and MRI in CD.

Search strategy and study selection

A comprehensive literature search was undertaken to identify all relevant citations. The electronic search strategy involved keyword searches of MEDLINE and EMBASE. The electronic search was supplemented by manually reviewing the reference lists of included studies as well as relevant review articles. The literature search comprised the period from January 1994 to December 2010 and the following search criteria were used (all fields): (‘Crohn Disease’ OR ‘Crohn’s’ OR ‘inflammatory bowel disease’) AND (‘ultrasound’ OR ‘ultrasonography’ OR ‘sonography’ OR ‘computed tomography’ OR ‘CT scan’, OR ‘Helical CT’, OR ‘CT enterography’ OR ‘magnetic resonance’ OR ‘magnetic resonance imaging’ OR ‘MRI’ OR ‘MR’). References from the articles so selected were also examined in search of additional studies meeting inclusion criteria.

Five reviewers (BM, TR, RB, SQ and JR) independently assessed the eligibility of the articles for inclusion. The search yielded a total of 1408 articles, of which 169 were found to be eligible and retrieved as full-text articles for conspicuous analysis (Figure 1).

Figure 1.

 Flow chart for the selection of the studies included in this systematic revision.

The final selection of the articles was performed according to the following criteria: (i) 15 or more patients were included; (ii) US and/or CT and/or MRI were used to diagnose CD, for assessment of disease extension and/or activity, or for assessing CD-related complications; (iii) Adequate reference standard, including ileocolonoscopy, capsule endoscopy, enteroscopy or surgical or pathological findings for evaluating small and large bowel, (iv) Prospective design; and (v) Data reported to allow calculation of sensitivity, specificity, accuracy or correlation values (in the case of disease severity). Articles published in any language were included. All studies fulfilling the selection criteria were included in the systematic review, without performing any additional formal quality assessment.5–9 Disagreements between the reviewers regarding study inclusion were resolved by consensus of all authors. Following those criteria, 68 articles fulfilling all inclusion criteria were finally included in the analysis. For each of these studies, the following variables were extracted in a predefined data extraction form: author, publication year, number of patients included, population (adult or children), open or blinded design, gold standard [ileocolonoscopy, small-bowel capsule endoscopy (SBCE) or surgery], number of patients positive and negative for the variable examined, sensitivity and specificity. The mean sensitivity and specificity was calculated and expressed as weighted mean [and corresponding 95% confidence interval (CI)] to make allowances for the number of patients included in each study. Sensitivity and specificity estimates were calculated on per-patient and per-bowel-segment bases (when available), because information about the capability of a given test in localising disease and determining the extent of disease, in addition to its capability in diagnosing disease, can be obtained from segmental data. This separate calculation is relevant, as analysis on a per-patient basis probably leads to overestimation of sensitivity values, because any patient with disease is considered to have true-positive findings without considering whether the localisation of disease is correct or not.10 Technical details for optimal use of US, CT and MRI for the study of Crohn’s disease are provided in supplementary Data S1.

The evidence level (EL) and grade of recommendation (GR) were graded according to the Oxford Centre for Evidence-Based Medicine (http://www.cebm.net/index.aspx?o=1025).

Accuracy of imaging techniques in the diagnosis of CD

For this section, studies were considered only if they included patients with suspected CD; when both suspected and established CD were reported, the studies were only selected if accuracy results were available for suspected CD alone. Studies including only patients with established CD were excluded. A summary of the studies included in this section and its main characteristics is provided in Table 1.

Table 1.   Accuracy of cross-sectional imaging in the diagnosis of Crohn’s disease (CD) compared with endoscopy or surgical findings on a per-patient basis
AuthorPatients no. included/no. confirmedPopulationReference standardLocation evaluatedSensitivity (%)Specificity (%)
  1. CT, computed tomography; SBE, small bowel enteroclysis; SBCE, small bowel-capsule endoscopy; NA, not available.

  2. * Calculated by two observers.

Ultrasonography
 Bozkurt et al.1546/31AdultsEndoscopy, barium, surgerySmall and large bowel75100
 Hollerbach et al.12168/69AdultsEndoscopy (ileocolonoscopy, gastroscopy), enteroclysis, barium enema, CT, surgerySmall and large bowel84NA
 Astergiano et al.13313/61AdultsEndoscopy, radiological proceduresSmall and large bowel8498
 Parente et al.14487/NAAdultsEndoscopy (ileocolonoscopy, gastroscopy, push-enteroscopy), SBE, barium enema, CT, surgerySmall and large bowel88NA
 Borthne et al.1115/NAChildrenColonoscopySmall and large bowel93NA
Magnetic resonance imaging
 Albert et al.1725/14AdultsIleocolonoscopy, SBCESmall bowel7780
 Pilleul et al.1815/6ChildrenIleocolonoscopySmall and large bowel83100
 Borthne et al.1120/NAChildrenColonoscopySmall and large bowel82100
 Horsthuis et al.1933/15ChildrenEndoscopy (ileocolonoscopy, gastroscopy), enteroclysis, clinicSmall and large bowel61–91*60–90*

Statement 1

  • (i)US is an accurate technique for the initial diagnostic workout of patients with suspected CD, with the advantages of being widely available, non-invasive and free of ionising radiation. [EL 2b, GR B]
  • (ii)As the accuracy of US is highly dependent on factors such as experience level of the examining physician and location and severity of the disease, in patients with high suspicion of CD negative US findings should be confirmed with other diagnostic techniques. [EL2b GR C]
  • (iii)Accuracy of CT for diagnosing of CD in patients with a suspected diagnosis has not been properly evaluated in prospective studies.
  • (iv)Magnetic resonance imaging is an accurate technique for the diagnosis of suspected CD, with the advantage over US of being less dependent on the examiner and disease location. [EL2b GR B]
  • (v)The accuracy of cross-sectional imaging techniques for the diagnosis of CD is dependent on disease severity and is inferior to endoscopy for detection of mild lesions. Cross-sectional imaging techniques allow identification of transmural CD and extraluminal lesions, which cannot be assessed by endoscopic methods. [EL 2b GR B]

Accuracy of US in the diagnosis of CD

Five studies including a total of 1029 patients with suspected CD were identified.11–15 The overall per-patient sensitivity of US for the diagnosis of CD calculated from all studies included (Table 1) was 85% (95% CI 83–87%). Overall per patient specificity derived from studies reporting this data were 98% (95% CI 95–99%).13, 15

Diagnosis of CD was based mainly on the measurement of bowel wall thickness in all studies. Most studies considered a bowel wall thickness of 4 mm or higher to be a pathological finding.11, 12, 14, 15 Other US findings considered to be significantly associated with a diagnosis of CD by the majority of authors include decreased compressibility of thickened bowel walls, narrowing of the lumen, conglomeration of loops and extramural lesions such as fistulas or abscesses.12, 14, 16

Disease location and disease activity are the main factors influencing the accuracy of US for the diagnosis of CD.11–14 When sensitivity was estimated based on disease location, the highest values were found for anatomic areas easily accessible by US, such as the terminal ileum and left colon, whereas the accuracy was fairly low for less accessible locations, such as the rectum and upper small bowel.14 Moderate heterogeneity was found among the sensitivities reported by the studies included.11–15 Differences in the resolution of ultrasonic units, the cut-off value for bowel wall thickness and the experience of sonographers may explain, at least in part, the differences in terms of accuracy between these studies.

Overall, the results show that US is an accurate technique to detect bowel wall abnormalities as an initial approach in patients with suspected CD, with the advantages of being a widely available, minimally invasive and a ionising radiation-free tool.

Accuracy of CT in the diagnosis of CD

No studies fulfilling all inclusion criteria were identified evaluating the accuracy of CT in the diagnosis of patients with suspected CD.

Accuracy of MRI in the diagnosis of CD

The accuracy of MRI in the diagnostic workout of patients with suscepted CD was evaluated in four studies fulfilling the inclusion criteria (Table 1).11, 17–19 Overall per-patient sensitivity and specificity of MRI for the diagnosis of CD, calculated from included studies, which are detailed in Table 1, were 78% (95% CI 67–84%) and 85% (95% CI 76–90%) respectively.

As with US, in addition to wall thickness, consideration of other changes that are associated with disease activity such as wall enhancement after injection of MRI contrast, and presence of oedema, increase the sensitivity of MRI for diagnosing CD.17–19 Some technical aspects, in particular distension of the bowel and use of a luminal contrast, may affect the accuracy of MRI for assessing changes associated with active disease such as wall thickening and enhancement of bowel wall after MRI contrast administration.20

Some prospective studies have shown that SBCE may be more sensitive than MRI for establishing a diagnosis of small bowel CD.17 Comparisons of the diagnostic yield of SBCE and MRI enteroclysis in the detection of small bowel lesions show that in patients with CD, SBCE is able to detect significantly more inflammatory lesions in the proximal and middle part of the small bowel in comparison with MR enteroclysis.21 Nevertheless, SBCE and MRI could be complementary methods for diagnosing small bowel CD; some lesions missed by SBCE can be captured by MRI and vice versa.22

Comparison of US, CT and MRI in the diagnosis of CD

The only direct comparison of the diagnostic accuracy of cross-sectional imaging techniques including a reference standard was performed by Borthne et al.11 In this study, US and MRI were compared using ileocolonoscopy as reference standard in paediatric patients with suspected IBD. US showed higher sensitivity than MRI (93% vs. 82%), but the sample size was very small and most importantly, only the terminal ileum, the location with the highest accuracy for US, was examined. No studies were found comparing CT enteroclysis or enterography with US specifically in patients with suspected CD.

Horsthuis et al. assessed the accuracy of US, MRI and CT in the diagnosis of IBD, when compared with a predefined reference standard in a meta-analysis including 33 studies.10 Mean sensitivity estimates for the diagnosis of IBD (both CD and/or ulcerative colitis) on a per-patient basis were high and not significantly different between the three imaging modalities (90%, 93%, and 84% for US, MRI and CT respectively). Mean per-patient specificity estimates were 96% for US, 93% for MRI and 95% for CT. Mean per-bowel-segment sensitivity estimates were lower: 74% for US, 70% for MRI and 68% for CT. Mean per-bowel-segment specificity estimates were also similar: 93% for US, 94% for MRI and 90% for CT. CT proved to be significantly less sensitive and specific compared with MRI, which in conjunction with the need to minimise radiation exposure makes the use of US or MRI preferable.

Assessment of disease extension

Statement 2

  • (i)Assessment of disease extension in the small bowel should be based on radiological imaging techniques. MRI and US have a high diagnostic accuracy for assessment of disease extension. Selection between MRI and US should be based on the anatomical location to be explored, local expertise and availability. [EL 1b, RG A]
  • (ii)For the assessment of jejunal and ileal lesions MRI is preferred over US for its higher sensitivity particularly for jejunal lesions. [EL 2b, RG B]
  • (iii)Assessment of disease extension in the colon and terminal ileum should be based on endoscopy and completed with imaging techniques in cases of incomplete procedures. [EL 1b, RG A]
  • (iv)Ultrasonography and MRI can be used as imaging methods for disease extension in the terminal ileum and colon. Higher availability and tolerance may render US a preferred technique. [EL 1b, RG A].
  • (v)Indirect evidence suggests a similar diagnostic accuracy for CT, but radiation exposure is a limitation for repeated use of this technique. [EL 5, RG D]

Location and extension of active CD with US

Eight studies compared the accuracy of US with other imaging techniques and/or endoscopy and/or surgery in localising CD lesions (Table 2).14, 23–29 The sensitivity of US for diagnosis of location ranged from 74% to 96%, with specificity ranging between 80% and 100%. Pooled results of these studies show a sensitivity of 86% (95% CI 83–88%) and specificity of 94% (95% CI 93–95%).

Table 2.   Accuracy of ultrasonography, computed tomography and magnetic resonance imaging in the assessment of disease extension in Crohn’s disease
AuthorPatients with CD no. total/no. activePopulationReference standardLocation evaluatedSensitivity (%)Specificity (%)
  1. CT, computed tomography; NA, not available.

Ultrasonography
 Maconi et al.24115/NAAdultsIleocolonoscopy, barium contrast studiesTerminal ileum and colon8994
 Reimund et al.2848/47AdultsIleocolonoscopy, barium contrast studiesTerminal ileum and colon8367
 Bru et al.2338/NAAdultsIleocolonoscopyTerminal ileum and colon8387
 Parente et al.29296/NAAdultsIleocolonoscopy, barium contrast studies (small bowel and barium enema)Terminal ileum and colon9397
 Parente et al.14273/NAAdultsEndoscopy, barium contrast studies (small bowel and barium enema), CT, surgeryTerminal ileum and colon7795
 Pascu et al.2737/NAAdultsIleocolonoscopyTerminal ileum and colon7497
 Parente et al.26102/NAAdultsIleocolonoscopy, barium enteroclysisSmall bowel9698
 Martinez et al.2530 (segments 119/53)AdultsEndoscopy, barium contrast studies (small bowel and barium enema), CTSmall bowel and colon9198
Computed tomography
 Fiorino et al.3044/28AdultsIleocolonoscopy, surgerySmall bowel and colon8888
Magnetic resonance imaging
 Low et al.3128/25AdultsBarium studies, high resolution endoscopy, surgerySmall bowel and colon86NA
 Pascu et al.2737/NAAdultsIleocolonoscopyColon and terminal ileum3890
 Albert et al.1752/30AdultsCapsule endoscopySmall bowel85100
 Tillack et al.2219/18AdultsCapsule endoscopySmall bowel7891
 Fiorino et al.3044/28AdultsIleocolonoscopy, surgerySmall bowel and colon8888

The addition of an oral non-absorbable solution (bowel hydrosonography) results in an increase in the sensitivity of US for the detection of segments with active disease.26 Moreover, the use of a luminal contrast in the colon (hydrocolonic sonography) provides a high accuracy for assessing colonic CD lesions.23

Location and extension of active CD with CT

The accuracy of CT for assessment of CD extension has been evaluated in only one prospective study with adequate reference standard, showing a high sensitivity (88%) and specificity (88%). However, the sensitivity for detection of lesions in colonic segments, particularly in the ascending colon (38%) was significantly lower than for the ileum (81%), which may be related to the lack of endoluminal contrast.30

Location and extension of active CD with MRI

Three studies assessed disease extension in the colon and ileum27, 30, 31 obtaining a sensitivity of 38–88% and specificity of 88–90%. In two other studies evaluating the accuracy of MRI only in the proximal small bowel, the technique correctly categorised disease extension in most patients.17, 22 The sensitivity of MRI for evaluating the extent of CD lesions in the small bowel was 74% (95% CI 68–80%) and the specificity 91% (95% CI 86–95%)17, 22 (Table 2).

Comparison of different cross-sectional imaging techniques for assessment of disease extension in CD

Ultrasonography seems to have a superior overall accuracy for the detection of disease localised in the terminal ileum and colon, except for the rectum and MRI has superior accuracy compared with US for the detection of lesions in the jejunum and more proximal ileum (89% vs. 73%).27 Direct comparison of CT and MRI for assessment of location and extension of inflammatory lesions demonstrated a similar diagnostic accuracy.30

Assessment of disease activity and severity

Statement 3

  • (i)Ultrasonography has a high diagnostic accuracy for assessment of disease activity in the terminal ileum and colon [EL 1b, RG A]. MRI may achieve a similar sensitivity if adequate luminal distension is achieved. [EL 1b, RG A]
  • (ii)Computed tomography can also be used to assess activity in the terminal ileum as accuracy is similar to other diagnostic techniques for this location [EL 1b, RG A]. Information is insufficient for determining accuracy of CT for colonic disease.
  • (iii)Ultrasonography, MRI and CT have a higher accuracy for assessing disease activity in terminal ileum than barium contrast studies. [EL 1b, RG A]
  • (iv)As a result of lack of radiation US or MRI should be preferred over CT for evaluation of disease activity and severity, particularly in young patients. [EL 5, RG D]

Assessment of disease activity

Assessment of disease activity with US.  Six studies assessed the presence of active disease in a total of 207 CD patients (Table 3).23, 28, 32–35 The sensitivity of US for the detection of disease activity ranged from 63% to 100%, with specificity in the range of 77–100%. Overall sensitivity was 85% (95% CI 79–89%) and specificity 91% (95% CI 87–95%).

Table 3.   Accuracy of ultrasonography, computed tomography and magnetic resonance imaging in the assessment of disease activity in Crohn’s disease
AuthorPatients with CD no. total/no. activePopulationReference standardLocation evaluatedSensitivity (%)Specificity (%)
  1. NA, not available; PET, positron emission tomography.

Ultrasonography
 Pradel et al.3519/18AdultsIleocolonoscopy, barium contrast studiesTerminal ileum and colon6390
 Reimund et al.2848/47AdultsIleocolonoscopy, barium contrast studiesTerminal ileum and colon96NA
 Bru et al.2338/NAAdultsIleocolonoscopyTerminal ileum and colon10082
 Miao et al.3330/23AdultsIleocolonoscopy, barium contrast studies, surgeryTerminal ileum and colon87100
 Bremner et al.3225/NAChildrenIleocolonoscopy, barium contrast studiesTerminal ileum and colon4898
 Migaleddu et al.3447/31AdultsIleocolonoscopy, histologyTerminal ileum and colon90 (wall thickness)
90 (colour Doppler)
93 (contrast-enhanced US)
93 (wall thickness)
93 (colour Doppler)
94 (contrast-enhanced US)
Computed tomography
 Low et al.4226/NA
(segments 128/65)
AdultsEndoscopy (ileocolonoscopy, gastroscopy), enteroclysis, barium enema, surgerySmall and large bowel60–6590–92
 Hassan et al.3630/30AdultsIleocolonoscopyTerminal ileum87100
 Wold et al.3923/13AdultsIleoscolonoscopyTerminal ileum7790
 Andersen et al.4115/NAAdultsColonoscopyColon7550
 Solem et al.3841/21AdultsConsensus panelSmall bowel8289
 Lee et al.3730/18AdultsIleoscolonoscopyTerminal ileum8980
 Siddiki et al.4033/22AdultsIleocolonoscopyTerminal ileum9588
 Fiorino et al.3044/28AdultsIleocolonoscopy, surgerySmall bowel and colon86–9091–100
Magnetic resonance imaging
 Low et al.4226/NA (segments 193/65)AdultsEndoscopy (ileocolonoscopy, gastroscopy), enteroclysis, barium enema, surgerySmall and large bowel80–8591
 Koh et al.4530/7 (segments 124/30)AdultsIleocolonoscopyColon and terminal ileum91
59
71
93
 Miao et al.3330/23AdultsIleocolonoscopyColon and terminal ileum8771
 Neurath et al.4451/51 (segments 139/42)AdultsIleocolonoscopy, PETColon and terminal ileum6793
 Ochsenkühn et al.4325/18AdultsIleocolonoscopyTerminal Ileum8885
 Pilleul et al.1862/23ChildrenIleocolonoscopyColon and terminal ileum7846
 Schreyer et al.4730/29 (segments 161/49)AdultsIleocolonoscopyColon and terminal ileum5598
 Florie et al.4831/21AdultsIleocolonoscopyColon and terminal ileum9355
 vanGemert-Horsthuis et al.4920/16AdultsIleocolonoscopyColon and terminal ileum100100
 Lee et al.3730/18AdultsIleocolonoscopyTerminal ileum83100
 Rimola et al.4650/35 (segments 218/83)AdultsIleocolonoscopyColon and terminal ileum8189
 Siddiki et al.4030/21AdultsIleocolonoscopyTerminal ileum9167
 Fiorino et al.3044/28AdultsIleocolonoscopy, surgerySmall bowel and colon81–9091–95
 Oussalah et al.5061/NA (segments 221/NA)AdultsIleocolonoscopyColon and terminal ileum5884
 Hyun et al.5130/20 230/64AdultsIleocolonoscopy, capsule endoscopySmall bowel
Colon
86
62
93
95
 Rimola et al.5248/29 (segments 258/115)AdultsIleocolonoscopySmall bowel
Colon
8787

Three studies used Doppler US for the detection of active disease, showing that wall thickness and the angiographic vascularisation pattern (Figure 2) are useful for the detection of active disease.16, 25, 34 To date, one study evaluated the accuracy of contrast-enhanced US for assessment of activity in CD, showing that the technique has a high sensitivity [93%, (95% CI 88–97%)] and specificity [94%, (95% CI 88–97%)].34 The comparison of the diagnostic accuracy of conventional US, Doppler US and contrast-enhanced US for assessment of disease activity, showed that the sensitivity of the three modalities of examination (94%, 94% and 94%) and specificity (97%, 97% and 97%) were virtually identical.34

Figure 2.

 Inflammatory changes in active Crohn’s disease assessed by US. Abdominal US shows bowel wall thickening with moderate positive vascularity on Doppler. Surrounding perienteric fat has an increased echogenicity because of inflammatory changes.

Assessment of disease activity with CT.  The diagnostic value of CT for distinguishing quiescent from active CD affecting only the terminal ileum was determined in eight studies (Table 3),30, 36–42 including a total of 216 patients with CD. The sensitivity of CT for the detection of disease activity in the terminal ileum ranged from 60% to 95%, with specificity ranging from 50% to 100%. Overall sensitivity was 81% (95% CI 77–86%) and specificity 88% (95% CI 82–91%). The lowest sensitivity for the detection of colonic lesions was obtained in the study not using luminal contrast.30

Assessment of disease activity with MRI.  Sixteen studies have evaluated the accuracy of MRI in the assessment of activity in the terminal ileum and/or the colon (Table 3).18, 30, 33, 37, 40, 42–52 Six of these studies18, 30, 33, 45, 48, 49 provide results on a per patient basis including a total of 217 patients, 11 studies provide results on a per segment basis,37, 40, 42–47, 50–52 in three of these only one segment, the terminal ileum, was studied;37, 40, 43 a total of 1629 segments were assessed in all these studies. Pooled results of studies determining accuracy of MRI for assessment of disease activity on a per patient basis show a sensitivity of 80% (95% CI 77–83%) and specificity of 82% (95% CI 78–85%) whereas pooled results of studies reporting on a per segment basis showed a sensitivity of 70% (95% CI 67–73%) and specificity of 89% (95% CI 93–96%). In general studies using luminal negative contrast show a higher accuracy for assessment of disease activity (Figure 3), and one study assessing the value of diffusion weighted sequences showed also a low sensitivity for the detection of colonic CD lesions in the absence of luminal contrast.50

Figure 3.

 Active Crohn’s disease in the terminal ileum evaluated by MRI. Coronal T1 with fat-suppressed MR image obtained after intravenous contrast injection shows mural hyperenhancement and mural thickening on the distal ileum (arrow). Prominent engorged vasa recta in the mesentery (comb sign) is also identified (arrowheads).

Comparison of different imaging techniques for detection of activity in CD.  Four studies compared the sensitivities and specificities for CT and MRI.30, 37, 40, 42 Three studies found a similar diagnostic accuracy for both techniques and one study using helical CT (the other three studies used multidetector CT) showed a higher sensitivity and specificity for MRI.42 Two studies compared the diagnostic accuracy of US and MRI to assess the presence of activity in the colon and terminal ileum.27, 33 In these studies, the diagnostic accuracy of US was higher than that of MRI. Again, studies reporting low sensitivity for MRI are those that did not use luminal distension producing more false negative results.

Assessment of disease severity

Statement 4

  • (i)A high correlation exists between the severity of intestinal lesions assessed by endoscopy and the intensity of US, MRI or CT changes. [EL 2b, RG B]
  • (ii)A weak correlation exists between findings of cross-sectional imaging techniques and clinical activity indexes or biomarkers. [EL 1b, RG A]
  • (iii)Ultrasonography, MRI or CT can be used in clinical practice for the assessment of disease severity. [EL 1b, RG A]

Assessment of disease severity with US.  Twelve studies evaluated the relationship between US findings and disease severity assessed byendoscopy, small bowel enterography, clinical indexes of disease activity (Van Hess and CDAI) and biomarkers (C-reactive protein, platelet count, erythrocyte sedimentation rate) (Table 4).14, 16, 23–25, 27, 29, 32, 34, 53–55 Two studies found that bowel wall thickening was related to the severity of inflammation, as assessed by endoscopic examination and histopathological analysis.32, 55 Contrast-enhanced US could classify severity significantly better than Doppler-US signal and measurement of mural thickening (< 0.001).53 In general, the magnitude of US changes have a high correlation with endoscopic and histological magnitude of alterations16, 27, 53–55 and a weak correlation with indexes of clinical activity and biomarkers.14, 23–25, 27, 29, 32, 54

Table 4.   Efficacy of ultrasonography, computed tomography and magnetic resonance imaging in the assessment of disease severity in Crohn’s disease
AuthorPatients with CD no. total/no. activePopulationReference standardLocation evaluatedResults
  1. CDAI, CD Activity Index; CRP, C-reactive protein; CT, computed tomography; NA, not available.

Ultrasonography
 Maconi et al.24115/NAAdultsIleocolonoscopy, barium contrast studiesTerminal ileum and colonLow correlation between US and CDAI or biological markers
 Futagami et al.5455/NAAdultsIleocolonoscopy, barium contrast studiesSmall bowel and colonCorrelation between ultrasonographic activity index and endoscopic or barium contrast studies r = 0.62
Correlation with CDAI r = 0.2 and biomarkers r = 0.08
 Bru et al.2338/NAAdultsIleocolonoscopyTerminal ileum and colonCorrelations between hydrocolonic sonographic activity index and Van Hess r = 0.72
Correlation with biomarkers r = 0.44 and r = 0.55
 Haber et al.5526/NA (segments 164/104)ChildrenIleocolonoscopy, histologyTerminal ileum and colonCorrelations between US and endoscopic and histological findings (P < 0.0001)
 Parente et al.29296/NA Ileocolonoscopy, barium contrast studies (small bowel and barium enema)Terminal ileum and colonNo significant correlation between US and CDAI (r < 0.6)
 Parente et al.14487/NA Endoscopy, barium contrast studies (small bowel and barium enema), CT, surgeryTerminal ileum and colonNo correlation between US and CDAI r = 0.29 or CRP r = 0.17
 Pascu et al.2737/NA IleocolonoscopyTerminal ileum and colonCorrelations between US and endoscopic activity index r = 0.83
No significant correlation with CDAI or CRP
 Neye et al.1622/NA (segments 126/70) Ileocolonoscopy, barium contrast studies, surgeryTerminal ileum and colonHigh concordance of power Doppler US and endoscopy; k coefficient 0.71–0.91
 Bremner et al.3225/NAChildrenIleocolonoscopy, barium contrast studiesTerminal ileum and colonNo significant correlation with superior mesenteric artery Doppler and disease severity
 Ripolles et al.5361/53AdultsIleocolonoscopyTerminal ileum and colonCorrelation between US (contrast-enhanced) and endoscopic severity P < 0.001
 Martinez et al.2530/NA (segments 119/53)AdultsEndoscopy, barium contrast studies (small bowel and barium enema), CTSmall bowel and colonSignificant correlation between bowel wall thickness and CDAI and CRP (P < 0.01)
 Migaleddu et al.3447/31AdultsIleocolonoscopy, histologyTerminal ileum and colonLinear correlation coefficient for contrast-enhanced US, baseline US, and colour Doppler US vs. CDAI 0.74, 0.68 and 0.73, respectively
Computed tomography
 Hassan et al.3630/30AdultsIleocolonoscopyTerminal ileumCorrelation with CDAI (r = 0.81)
 Andersen et al.4115/NAAdultsIleocolonoscopy,ColonCorrelation enhancement – endoscopic inflammation (r = 0.72)
Magnetic resonance imaging
 Pascu et al.2737/NAAdultsIleocolonoscopyColon and terminal ileumCorrelation between MRI index and endoscopic activity index r = 0.344 (P = 0.007)
 Florie et al.4831/21 (segments 141/NA)AdultsIleocolonoscopyColon and terminal ileumCorrelation between MRI and endoscopic severity: 0.59 (observer 1) – 0.53 (observer 2) (P < 0.001)
 Schreyer et al.4730/29AdultsIleocolonoscopyColon and terminal ileumDistinction between mild-moderate and severe
Sensitivity 69%, Specificity 99%
 van Gemert-Horsthuis et al.4920/15Adults yIleocolonoscopColon and terminal ileumLack of correlation between MRI severity and endoscopic severity: r = 0.4 (P = 0.09)
 Tillack et al.2219/18 (segments 40/NA)AdultsCapsule endoscopySmall bowelDistinction between mild-moderate and severe:
Sensitivity 58%, Specificity 77%
 Rimola et al.4629/48 (segments 115/258)AdultsIleocolonoscopyColon and terminal ileumDistinction between mild-moderate and severe
Sensitivity 91%, Specificity 95%
 Oussalah et al.5061/NA (segments 211/NA)AdultsIleocolonoscopyColon and terminal ileumCorrelation between MRI index and endoscopic severity index r = 0.659 (P < 0.001)
 Rimola et al.5248/NA (segments 158/115)AdultsIleocolonoscopyColon and terminal ileumDistinction between mild-moderate and severe
Sensitivity 92%, Specificity 92%
Correlation between MRI index and endoscopic severity index r = 0.84 (P < 0.001)
 Hyun et al.5130/20AdultsIleocolonoscopy, double balloon enteroscopySmall bowel and colonCorrelation between MRI index and endoscopic severity index r = 0.85 (P < 0.001)

Assessment of disease severity with CT.  Two studies evaluated the value of CT for assessment of disease severity.36, 41 One study found a significant correlation between the degree of enhancement and the severity of CD assessed by CDAI36 that was not confirmed in the other study.41 In the latter, significant correlations were observed between the intensity of various CT changes including wall thickening, enhancement and presence lymphadenopathy and the severity of endoscopic lesions (Table 4).

Assessment of disease severity with MRI.  Nine studies evaluated the accuracy of MRI in the assessment of disease severity in the colon and distal ileum,22, 27, 46–52 showing a good correlation with endoscopic findings, that was lower when the intestinal lumen was not distended50, 51 (Table 4). Three studies derived an index of activity, one based on a qualitative evaluation of MRI findings including contrast enhancement, oedema, wall thickening, ulcer (Figure 4), presence of layered pattern and diffusion hyperintensity,50 and two others by the same group of investigators calculating a quantitative index based on the first four variables.46, 52 The later studies showed a higher correlation with the severity of CDEIS (r = 0.80 and r = 0.84) than the first one (r = 0.54), although in addition to differences in derivation methodology, technical aspects may also have influenced the results. A formal validation study of a radiological-based index of severity has only been performed for MRI.52

Figure 4.

 Ulcerative lesions of large bowel in a patient with Crohn’s disease. Ulcers appear at MRI (arrowheads in ‘a’) as a deep depressions on the mucosal surface of the thickened bowel wall. Colonoscopy (b) revealed deep ulcers in the colon surrounded by a markedly inflamed mucosa.

One study evaluated the diagnostic accuracy of MRI for assessing the severity of small bowel CD using SBCE as reference standard for evaluation of a total of 52 segments. MRI underestimated severity in 14% and revealed more severe pathology in 11.5%; in the remaining segments (75.5%), there was agreement between the two techniques.22 The high accuracy of MRI for assessment of upper small bowel lesions is confirmed in a study using double balloon enteroscopy showing a correlation of 0.88 between an endoscopic and an MRI index of activity based on the same parameters used in the above mentioned studies.51 The majority of studies found no correlation between MRI findings and indices of clinical activity or biomarkers,18, 27, 43, 44, 48, 49 and in three studies, this correlation was weak.46, 50, 56

Comparison of different imaging techniques for assessment of severity in CD.  One single study compared the accuracy of MRI without luminal contrast and US in the evaluation of disease severity in the colon and terminal ileum using ileocolonoscopy as reference standard.27 US showed a better correlation with endoscopic findings (r > 0.8) than MRI (r > 0.50).

Assessment of postsurgical recurrence

Statement 5

  • (i)Ultrasonography, CT and MRI are valid alternatives to ileocolonoscopy for assessing disease recurrence in patients who have undergone ileocolic resection. The degree of alterations in radiological findings parallels the severity of endoscopic lesions.[EL 2b, RG B]

The summary of studies assessing the value of US, CT and MRI for assessment of post-operative disease recurrence is shown in Table 5. Four studies evaluated the accuracy of US for assessment of postsurgical recurrence,57–60 two using oral contrast;57, 59 all used endoscopy as reference standard and the Rutgeerts’ score to categorise severity. All studies showed a high accuracy of US for the diagnosis of postsurgical recurrence in CD, detecting almost all cases of severe or complicated recurrence, as well as a high sensitivity and specificity in differentiating mild from severe recurrence.57–60 In one study, a direct comparison between US with and without oral contrast was performed, showing a mild increase in the sensitivity after giving oral contrast while specificity remained unchanged.59

Table 5.   Accuracy of ultrasonography, computed tomography and magnetic resonance in the assessment of postsurgical recurrence in Crohn’s disease (CD)
AuthorPatients with CD no. total/no. activePopulationRadiologic techniqueReference standardSensitivity (%)Specificity (%)
  1. NA, not available.

Rispo et al.5845/24AdultsTransabdominal USIleocolonoscopy79
Severe disease: 94
95
Severe disease: 100
Sailer et al.6330/27AdultsMR enteroclysisIleo-colonic anastomosisNA
Correlation with endoscopic severity r = 0.67 (P < 0.01)
Castiglione et al.5922/24AdultsTransabdominal USIleocolonoscopy7794
Oral contrast Abdominal US8294
Minordi et al.6134/32AdultsCT EnteroclysisIleocolonoscopy97100
Calabrese et al.5772/63AdultsSmall intestine oral contrast USIleocolonoscopy9220
Paredes et al.6026/33AdultsTransabdominal USIleocolonoscopy  
 Wall thickness8077
 Doppler US8067
 1 or 2 of the above8667

One study evaluated CT for detecting anastomotic recurrence, showing high sensitivity, specificity and accuracy in the evaluation of relapse of ileocolic anastomosis.61

Five published studies assessing the accuracy of MRI included operated patients,33, 45, 48, 62, 63 but only one allowed objective analysis of the results in this subgroup.63 The correlation for the assessment of activity between endoscopy (Rutgeerts’ score) and MRI and differentiation between mild and severe lesions was considered very high (k coefficient >0.8) (Table 5).

Limitations of selected studies on disease extension, activity and severity

The main limitation for the assessment of the diagnostic accuracy of imaging techniques in the diagnosis of small bowel CD is the lack of an established reference standard. SBCE is currently the only technique that allows direct visualisation of the whole small bowel. However, SBCE also has a series of limitations including the risk of retention because of strictures,17 the lack of validated diagnostic criteria for interpretation of the significance and severity of the lesions detected in the small bowel, and the fact that evaluation is limited to the surface of the intestinal mucosa, which may explain an apparent ‘over-rating’ of disease activity/severity by MRI in some patients.22

Another important limitation is the small sample size in the majority of studies and that most studies evaluated only the presence of active disease, without attempting to differentiate between mild and severe lesions, which is crucial for therapeutic decision-making.

In studies evaluating MRI, heterogeneous results may be related to the type of units (3 T vs. 1.5 T) and most importantly the use of luminal distension. Reporting the results on a per-patient or per-segment basis also has a significant effect in the assessment of diagnostic accuracy, which is always lower in the latter.

None of the studies analysed the influence of concomitant medications as a covariate in the models to correlate radiological findings and disease activity.8 This omission may be of some importance, as, for example, steroids have been shown to reduce the diagnostic accuracy of scintigraphy.64

Diagnosis of complications

Colonoscopy and barium administration have been the principal means of evaluating CD patients because of their ability to depict bowel mucosa alterations and stenosis. However, these techniques are unable to show either the transmural inflammation or extraluminal complications that are characteristic of CD. The ability to demonstrate directly the bowel wall, the adjacent abdominal organs, mesentery and retroperitoneum makes cross-sectional imaging techniques important diagnostic modalities for assessment of complications in CD.

Detection of intra-abdominal fistulas

Statement 6

  • (i)Ultrasonography, CT and MRI have a high sensitivity and specificity for the diagnosis of intra-abdominal fistulas, with similar diagnostic accuracies. [EL 2b, GR B]
  • (ii)Diagnostic accuracy of cross-sectional imaging techniques (US, CT and MR) for diagnosis of fistulas is higher than that of SBFT and should be preferred over the latter. [EL 2b, GR B]
  • (iii)Combinations SBFT with a cross-sectional imaging modality may increase the diagnostic accuracy over either technique alone. [EL 2b, GR B]
  • (iv)As a result of lack of radiation US or MR should be the preferred over CT for the detection of complications. Selection between MR and US will depend on local expertise and availability. [EL 5 GR D]

The diagnostic accuracy of US for the detection of fistulas (Figure 5) was assessed in four studies that included a total of 99 lesions affecting the small intestine or colon in 216 patients (Table 6).25, 65–67 The sensitivity of US for the diagnosis of fistulising lesions ranged from 67% to 87%, with specificity in the range of 90% to 100%. Pooled results of the three studies showed a sensitivity of 74% (95% CI 67–79%) and specificity of 95% (95% CI 91–97%).

Figure 5.

 Fistulising Crohn’s disease. Abdominal ultrasonography shows a lineal communication between two thickened small bowel loops (SB) which contains air (arrow), corresponding to an entero-enteric fistula.

Table 6.   Accuracy of ultrasonography, computed tomography and magnetic resonance imaging in the assessment of complications in Crohn’s disease (CD)
AuthorPatients with CDReference standardComplication (n)Sensitivity (%)Specificity (%)
  1. NA, not available.

Ultrasonography
 Maconi et al.65128SurgeryFistula (56)7196
Abscess (26)8093
 Maconi et al.7098Endoscopy, barium radiology, computed tomography, surgeryStenosis (40)7593
Abscess (6)8394
 Gasche et al.6633SurgeryStenosis (22)100901
Fístula (23)8790
Abscess (9)10092
 Kohn et al.6725SurgeryStenosis (16)7589
Fistula (3)6795
 Martínez et al.2530Endoscopy, barium radiology, surgeryFístula (17)82100
Computed tomography
 Turetschek et al.6928Endoscopy, surgeryStenosis (21)90100
Fistula (14)100100
Abscess (3)100100
 Maconi et al.6561SurgeryFistula (38)6891
Abscess (22)8695
 Hassan et al.3630EndoscopyStenosis (20)NANA
Fistula (3)
Abscess (5)
 Wold et al.3923Surgery, clinical examinationFistula (6)100100
Abscess (4)100100
 Voderholzer et al.7256Capsuleendoscopy, endoscopyStenosis (16)94100
 Hara et al.6817Ileoscopy, capsule endoscopy, SBFTFistula (2)NANA
 Solem et al.3841Use of a consensus clinical diagnosis as the criterion standardStenosis (8)88100
 Lee et al.3730Endoscopy, SBFT, surgeryFistula (11)100100
Abscess (5)100100
 Fiorino et al.3044Endoscopy, surgeryStenosis (14)85100
Fistula (5)20100
Magnetic resonance imaging
 Magnano et al.7122EndoscopyStenosis (8)8792
Abscess (1)NANA
 van Gemert-Horsthuis et al.4920EndoscopyStenosis (7)8685
 Pilleul et al.1847US, surgeryStenosis (8)100100
Fistula (9)100100
 Florie et al.4831EndoscopyStenosis, fistula, abscess (8)7591
 Maccioni et al.5659Endoscopy, CT, US, barium radiology, surgeryStenosis (22)91100
Fistula (9)78100
Abscess (4)8693
 Negaard et al.7435EndoscopyStenosis (9)8693
 Tillack et al.2219 pacients (52 segments)Capsule endoscopy, endoscopyStenosis (22)8293
 Martínez et al.2530Endoscopy, barium radiology, surgeryFistula (17)7192
 Lee et al.3730Endoscopy, barium radiology, surgeryFistula (11)100100
Abscess (5)100100
 Fiorino et al.3044Endoscopy, surgeryStenosis (14)9290
Fistula (5)4094

The diagnostic value of CT for diagnosing fistulas (Figure 6) was determined in seven studies that included a total of 79 lesions in 233 patients (Table 6).30, 36, 37, 39, 65, 68, 69 Five of these studies provided an appropriate reference standard for evaluation of extraenteric lesions such as surgery.30, 37, 39, 65, 69 Sensitivity ranged from 20% to 100% and specificity from 91% to 100%. Pooled results of the five studies with adequate reference standard showed a sensitivity of CT for the diagnosis of fistulas of 70% (95% CI 64–76%) and specificity of 97% (95% CI 94–99%).

Figure 6.

 Fistulising Crohn’s disease. Abdominal CT scan in axial plane after iv contrast administration depicts a thickened ileal loop with ileo-cecal fistula and surrounding perienteric inflammatory changes (arrow). Posterior to the fistula a small abscess is observed (arrowhead).

The diagnostic value of MRI for diagnosing intra-abdominal fistulas was determined in five studies that included a total of 51 lesions in 210 patients (Table 6).18, 25, 30, 37, 56 As in US and CT studies, the majority of lesions corresponded to enteroenteric fistulas. Four studies used a combination of diagnostic techniques, physical examination (enterocutaneous fistulas) and surgery as reference standard. In one study, no reference standard was provided.49 The sensitivity of MRI for fistulising lesions ranged from 40% to 100% and specificity from 93% to 100%. Pooled results of the studies with adequate reference standard showed a sensitivity of MRI for the diagnosis of fistulas of 76% (95% CI 71–82%) and specificity of 96% (95% CI 92–98%).

Various studies have compared the performance of different cross-sectional imaging modalities and small bowel enteroclysis. The accuracy of small bowel enteroclysis and US for the diagnosis of internal fistulas and abscesses complicating CD was compared in a study using surgery as reference standard, demonstrating that both techniques have the same accuracy for the detection of internal fistulas (85%).65 In the same study, CT was carried out in a subgroup of patients with suspicion of septic complications. In these patients, CT showed a sensitivity of 68% and a specificity of 91%. US had a sensitivity and specificity of 87% and 91%, whereas SBFT had a sensitivity of 74% and a specificity of 91%.

For the detection of internal fistulas, the combination of small bowel enteroclysis and US significantly improved diagnostic accuracy (small bowel enteroclysis 84%, US 85%, combination 91%). In the group of patients also undergoing CT examinations, the combination of US with small bowel enteroclysis and CT significantly improved sensitivity relative to CT or SBFT alone (97%, 68%, and 73% respectively), as well as overall accuracy.65

In another study using a combination of endoscopy, barium studies, CT and surgery as reference standard, 17 cases with enteroenteric fistulas were identified.25 US and MRI detected 14 (82%) and 12 (70%) fistulas respectively. The specificity and accuracy were 100% and 90% for US and 92% and 80% for MRI.

A number of studies fulfilling the selection criteria of the current systematic review have compared the diagnostic accuracy of CT and MRI for the detection of extraenteric complications. The study by Lee et al.37 showed that CT and MRI enterography have a higher accuracy than SBFT to identify extraenteric complications of CD, including fistulas, sinus tracts and abscesses; sensitivity values for the detection of extraenteric complications were significantly higher for CT and MR enterography (100% for both) than for SBFT (32–37%). CT and MR enterography were equally accurate for the identification of extraenteric complications (sensitivity 100% for both). Another study by Fiorino et al. observed also a similar accuracy of CT and MRI for the detection of fistulas, although this particular study reported the lowest sensitivity of both techniques for this type of lesions.30

A feasibility study to evaluate two CT enterography protocols (enterography and enteroclysis) in comparison with SBFT, terminal ileoscopy and surgery (in selected cases) showed a higher sensitivity of CT for the detection of extraenteric findings of CD (fistulas and sinus tracts, in addition to abscesses or phlegmons) compared with SBFT.39

Detection of abscesses

Statement 7

  • (i)Ultrasonography, CT and MRI have a high sensitivity for the diagnosis of intra-abdominal abscesses. Diagnostic accuracy of US is slightly lower than that of CT and MRI because of false positive cases. [EL 2b, GR B]
  • (ii)Systematic combination of cross-sectional diagnostic modalities does not significantly improve the diagnostic accuracy for the detection of intra-abdominal abscesses complicating CD, but CT or MRI may be used to confirm doubtful US lesions. [EL 5, GR D]
  • (iii)Cross-sectional techniques have a lower sensitivity for the detection of deep abscesses (e.g. retrogastric, deep pelvis). [EL 2b, GR B]

The value of US for the detection of abscesses has been assessed in three studies using surgery as reference standard, which included a total of 42 lesions in 242 patients (Table 6).65, 66, 70 The sensitivity of US for the diagnosis of abscesses ranged from 81% to 100%, with specificity in the range of 92% to 94%. Pooling the results of the three studies, the sensitivity was 84% (95% CI 79–88%) and specificity 93% (95% CI 89–95%). In the study by Gasche et al., intra-abdominal abscesses were correctly detected in 9/9 patients and excluded in 22/24 patients (sensitivity 100%, specificity 92%).66 Nevertheless, the apparently higher accuracy compared with other studies may be due, at least in part, to patient selection. It is widely recognised that certain anatomic areas are difficult to assess by US, in particular the stomach, the deep pelvic part of the sigmoid and the rectum. Accordingly, in this study, the authors did not attempt to detect complications in these areas. However, one of the three unrecognised fistulas was an ileosigmoid tract. This finding indicates that not only the distal sigmoid, but also the deep pelvic part of the ileum, may not be visualised by US.

The diagnostic value of CT for diagnosing intra-abdominal abscesses was determined in five studies that included a total of 39 lesions in 172 patients (Table 6).36, 37, 39, 65, 69 Four of these studies provided an appropriate reference standard for evaluation of extraenteric lesions,37, 39, 65, 69 and the other study used only endoscopy as reference standard.36 The sensitivity ranged from 87% to 100% and specificity from 95% to 100%. Pooling the results of the four studies with adequate reference standard, the sensitivity of CT for the diagnosis of abscesses was 84% (95 CI 78–90%) and specificity 97% (95% CI 94–99%).

Four studies using MRI reported on the presence of intra-abdominal abscesses, identifying 10 lesions in 109 cases (Table 6).37, 48, 56, 71 One study did not use an appropriate reference standard for the identification of extraenteric lesions;48 in the remaining studies, lesions were confirmed at surgery in the majority of cases (8/10). The results of studies with adequate reference standard show a sensitivity of MRI for the detection of abscesses of 86% (95% CI 79–91%) and a specificity of 93% (95% CI 88–97%).

A comparison of US and CT for the identification of abscesses in the study by Maconi et al.65 shows that abscesses were correctly detected in a high and similar proportion of cases by means of US (90.9%) and CT (86.4%), although accuracy was higher for CT (92%) than for US (87%) because of false positive results in US studies. Both methods missed only deep abscesses: five by US in the entire series of patients (three interloop, one mesenteric and one appendicular) and three by CT (two interloop and one mesenteric). The combination of CT and US did not significantly improve the diagnostic accuracy for detection of intra-abdominal abscesses complicating CD. Wold et al. demonstrated that CT was more sensitive in the detection of abscesses or phlegmons than SBFT.39 Among the four abscesses confirmed at surgery in 23 patients, three of the lesions were only demonstrated by CT.

Detection of stenosis

Statement 8

  • (i)Ultrasonography, CT and MRI have a high sensitivity and specificity for the diagnosis of stenosis affecting the large or small bowel. [EL 2b, GR B]
  • (ii)For US routine use of endoluminal contrast is not necessary. [EL 2b, GR C]
  • (iii)Diagnostic accuracy of MRI and CT is based on the use of luminal contrast. In partially obstructing stenosis enteroclysis may provide higher sensitivity than enterography for detection of lesions in the small bowel. [EL 2b, GR C]
  • (iv)Examination with a cross-sectional imaging modality is necessary before SBCE to rule out significant stenotic lesions. [EL 2b, GR C]

Ultrasonography has demonstrated a high diagnostic accuracy for the detection of small bowel stenosis in three studies, all using surgery as reference standard, identifying 78 stenotic lesions in 156 patients66, 67, 70 (Table 6). The sensitivity of US for the diagnosis of stenosis ranged from 74% to 100%, with specificity in the range of 89% to 93%. Pooling the results of the three studies, the sensitivity was 79% (95% CI 71–84%) and specificity 92% (95% CI 87–96%).66

Computed tomography enterography or enteroclysis has been shown to have a high diagnostic accuracy for the diagnosis of small bowel stenosis in five studies evaluating 79 stenoses of the small or large bowel in 199 patients (Table 6).30, 36, 38, 69, 72 In the four studies with adequate reference standard, the sensitivity of CT ranged from 85% to 93% and the specificity was 100%.30, 38, 69, 72 Pooling the results of the two studies, sensitivity was 89% (95% CI 83–94%) and specificity 99% (95% CI 97–100%).

Data derived from one of the CT studies suggest that use of a peroral contrast agent may have some limitations compared with nasojejunal intubation in the detection of small-bowel strictures, especially for partially obstructing strictures.73 Therefore, direct volume challenge via nasojejunal catheter is probably a preferred method for the diagnosis of strictures of the small bowel in patients with CD.

Eight studies report on the value of MRI for the detection of stenosis in CD, identifying 89 lesions, the majority in the small bowel, in 239 patients (Table 6).18, 22, 30, 48, 49, 56, 71, 74 One study identified seven patients with colonic stenoses at endoscopy, which were not assessed by MRI because this examination was applied only for assessment of the small bowel.37 MRI sensitivity for detection of stenoses ranged from 75% to 100% and specificity from 91% to 100%. Pooling the results of the seven studies with adequate reference standard, the sensitivity of MRI for diagnosis of stenosis was 89% (95% CI 84–92%) and specificity 94% (95% CI 90–96%).

One study comparing two methodologies of distension of the small bowel, enterography and enteroclysis, showed that bowel distension was also superior with enteroclysis, and that the latter had a higher sensitivity (86% vs. 100%) and specificity (93% vs. 100%) for the detection of stenoses.74 The direct comparison of CT and MRI for diagnosis of stenosis in the study by Fiorino et al. showed also a similar sensitivity (85% vs. 92%) and specificity (100% vs. 90%) for both techniques.30

Three studies22, 38, 72 indicate that small-bowel radiological imaging, preferably CT enterography of MR enterography, is needed prior to SBCE in suspected or known CD patients because of the high frequency of asymptomatic stenoses, and because these techniques can also detect extraenteric complications.

Limitations of selected studies on extraenteric CD complications

The high prevalence of intestinal complications in studies using surgery as the reference standard incorporates a selection bias. Fistulas, strictures and abscesses are present in a higher proportion of operated patients than in the general CD population. This may lead to overestimation of the specificity.

Comparisons of the diagnostic accuracy of various techniques across studies are also limited by factors related to the different definitions of lesions used. The definition of strictures has varied and includes the presence of severe luminal narrowing in regions of bowel wall thickening with or without prestenotic dilatation,66 identification of small bowel narrowing accompanied by prestenotic dilatation,30, 38 or a lesion causing narrowing of <1 cm.72 The definition of fistulas also varies between studies.65, 66, 69

It is important to establish a standard definition for each complication from a radiological point of view and for each of the available techniques. Such a definition would make it possible to reach more accurate overall conclusions from various studies, as we could then be confident that all the investigators assessed the same type of complication. In this sense, the most recent literature allows for these definitions for lesions detected with US and CT; however, standard definitions are more difficult to find for MRI.

Acknowledgements

Medical writing assistance was provided by Juliane Weber and Anna Mett, of inScience Communications, a WoltersKluwer business. This assistance and logistics for development of this consensus was supported by Abbott Laboratories. Declaration of personal interests: JP has served as a speaker and advisory board member for Abbott, MSD and Pfizer laboratories, and has received research grants from Abbott and MSD laboratories. Declaration of funding interests: None.

Ancillary