Key words abdomen, AREAS - STRUCTURES & SYSTEMS - fistula, THEMES - abscess - inflammation -
MR imaging
Introduction
The treatment of inflammatory bowel disease (IBD) has evolved from treating symptoms
to achieving mucosal healing. The “treat to target approach” demands
close monitoring of disease activity involving different methods [1 ]. For this reason, therapeutic decisions are based on
clinical parameters, additional laboratory and fecal markers, enterography results,
and endoscopic findings [2 ]
[3 ]. Ileo-colonoscopy is the first-line method to assess disease activity
and to monitor the development of malignancies in IBD patients. However, patients
are often reluctant to have a colonoscopy because of the invasiveness and pain.
Moreover, it is not feasible to repeat the procedure in short time intervals for
disease monitoring. Also, it represents an invasive procedure with procedure-related
risks such as bowel wall perforation, especially in IBD patients [4 ].
Magnetic resonance enterography (MRE) and transabdominal ultrasound (US) are widely
available for diagnosing and monitoring disease activity in IBD patients in
industrialized countries. Both methods are sometimes combined, e. g., in
complicated Crohn´s disease. However, the diagnostic value of each
individual method is unclear. In addition, there is suspected uncertainty of
inter-observer differences in the outcome of US [5 ].
On the other hand, US is cheap, easily accessible, and well tolerated compared to
MRE [6 ]
[7 ]. However, MRE
has a higher accuracy with respect to diagnosing abdominal abscesses and IBD
manifestations involving deeper pelvic loops or located proximal to the terminal
ileum [8 ]. Also, MRE shows high accuracy for
predicting the need for surgery in patients with Crohn´s disease and
strictures [9 ]. Although several studies point to
sufficient diagnostic accuracy of bowel US in comparison to MRE in IBD patients, it
is still not clear whether both methods should be combined in order to increase
diagnostic sensitivity and to improve disease course by guiding clinicians´
decisions [10 ].
We therefore aimed to compare the usefulness of US and MRE for assessing disease
extent and activity in the small and large bowel in patients with Crohn´s
disease and ulcerative colitis, and to assess the relevance for clinical decisions
even if endoscopy is not available.
Materials and Methods
We reviewed the medical records of all 232 patients who underwent MRE between January
2017 and May 2020. In this group, 124 patients were identified with IBD. Sixty-one
patients had undergone bowel US and MRE within three months. Six patients were
excluded due to a therapy switch between examinations. One patient was excluded due
to multiple abscesses on MRE. The remaining 54 patients were included in our
analysis. Our collective contained newly diagnosed patients with IBD and patients
with established disease in whom relapse was suspected. All US examinations were
performed according to the EFSUMB recommendations and clinical guidelines for
intestinal ultrasound in inflammatory bowel diseases [11 ]. Intestinal ultrasound was performed with high-end devices using
convex (1–6 MHz) and linear (2–9 and
6–15 MHz) probes (GE Health Care GmbH, Germany (Logiq E7-E9)). The
US examinations were performed by experienced (>4 months of practice)
physicians and cross-checked by the attending (>10 years of experience). The
activity of inflammation was determined by B-scan mode and Doppler mode. Therefore,
we used the Limberg score (including hypoechoic wall
thickening>4 mm, loss of wall stratification, and increase of
vascularity).
MRE examinations were performed either at 1.5 T or 3 T. Patients were
instructed to drink 1,500 ml of a 2% sorbitol solution during the
60 minutes before undergoing MRE. A 20-mg dose of hyoscine butylbromide
(Buscopan, Boehringer Ingelheim, Germany) was intravenously administered directly
before the start of the scan [12 ]. The imaging
protocol consisted of coronal and transverse T2-weighted imaging sequences and
transverse and coronal T1-weighted fat-suppressed T1-weighted imaging sequences
before and after intravenous contrast injection.
MRE examinations were evaluated in consensus by two radiologists (one radiologist
with>2 years of experience and one attending with>10 years of
experience). Bowel segments with suspicion of inflammation were defined by bowel
wall thickness>3 mm and increased vascularity detected by Doppler
signal on US or by contrast enhancement of the bowel wall on MRE.
The primary outcome was detection of bowel inflammation by US and MRE in IBD patients
validated against a construct reference standard [13 ].
Each patient’s record was reviewed by one gastroenterologist with>5
years of experience and one attending with>10 years of experience and the
suspicion of bowel inflammation was challenged by other established diagnostic
markers of disease activity (endoscopy or surgery with inflammation in histology,
CRP>5.0 mg/l, white blood cell (WBC)
count>11.0×109 /L, fecal calprotectin
concentration>40 μg/g, clinical symptoms, response
to steroid therapy). Then a consensus decision about the presence of bowel
inflammation at the time of enterography was made and used as the reference standard
([Supplementary table 1 ]).
In addition, correct identification of disease extent ((neo-) terminal Ileum vs.
colon) and identification of pathological findings such as fistulae, abscesses,
stenoses, or conglomerate tumors by US and MRE were compared. These secondary
outcomes were also validated against a construct reference standard based on a
consensus decision after reviewing each patient’s record.
Descriptive statistics of the baseline data are presented as medians [interquartile
range], or as percentages when appropriate. The sensitivity, specificity, accuracy,
positive predictive value [PPV], and negative predictive value [NPV] were calculated
with a 95% confidence interval using the retrospective final clinical
decision as the reference standard. Sensitivities were compared using
McNemar’s comparison of paired proportions. Differences in parameters were
accepted as statistically significant, when the probability of error for zero
hypothesis was less than 5%. All statistical analyses were performed by SPPS
version 27 for Mac OS (SPSS Inc., Chicago, IL, USA).
Results
A total of 54 patients with IBD (44 Crohn’s disease (CD), 5 ulcerative
colitis (UC) and 5 indeterminate colitis (IC)) were analyzed retrospectively. The
median time between US and MRE was 5 days (IQR 2.8–28.0 days). In our cohort
the median age was 29.5 years (IQR 21–47 years), and 38 patients
(70.4%) were female. Fourteen patients (26.0%) were diagnosed with
IBD for the first time. In the remaining 40 patients, the median disease duration
was 6.7 years (IQR 2.5–17.2 years).
43 out of 54 patients (76.8%) had not undergone previous surgery. Seven
patients had undergone prior ileocecal resection or right hemicolectomy resulting
in
a neo-terminal ileum, and two patients had undergone left hemicolectomy. One patient
had had a colectomy and another patient a proctocolectomy in the past. Endoscopic
findings with histology were available in 25 patients. Calprotectin values at the
time of enterography were available in 26 patients. Laboratory results including CRP
and blood count as well as information on the clinical course (fever, abdominal
pain, and diarrhea) were available in all 54 patients ([Table 1 ]).
Table 1 Patient characteristics in our study population of 54
IBD patients at the time of enterography.
Parameter
All patients n=54
Diagnosis
Crohn’s disease
44 (81.5%)
Ulcerative colitis
5 (9.3%)
Indeterminate colitis
5 (9.3%)
Sex
Female
38 (70.4%)
Male
16 (29.6%)
Age (median; IQR 25–75)
29.5, 21.0–47.3 years
Body mass index (median; IQR 25–75)
21; 19–24
New diagnosis
14 (26.0%)
Disease duration (median; IQR 25–75)
All patients
3.7; 0.1–10.4 years
Previous surgery
11 (20.4%)
Disease extent in patients with inflammation
N=42/54
Small bowel alone
18 (33.3%)
Colon alone
12 (22.2%)
Small bowel and colon
12 (22.2%)
Time between bowel US and MRE (median; IQR
25–75)
5; 2.8–28.0 days
Calprotectin (median, IQR 25–75)
952; 482–235 µg/g
Mayo score (if endoscopy is available)
N=25
Score 0
7 (28%)
Score 1
4 (16%)
Score 2
9 (36%)
Score 3
5 (20%)
According to the construct reference standard, 42 patients (78.0%) were
considered to have luminal bowel inflammation at the time of imaging. Eighteen
patients had isolated inflammation in the small bowel, compared to twelve patients
with isolated colonic inflammation. In the remaining 12 patients, the colon and
small bowel were affected simultaneously. Complications defined as abscesses,
conglomerate tumors, fistulae, or stenoses were found in 19 patients
(35.2%). 9 patients (16.7%) had more than one complication.
Clinical decisions in these 42 patients led to initiation of steroid treatment in
20
patients, antibiotic treatment in 5 patients, initiation, change or adaptation of
antibody treatment in 8 patients, and surgery or placement of percutaneous drainage
of an abscess in 6 patients. Three patients received other treatments
(5-aminosalicylic acid and azathioprine) ([Table
2 ]).
Table 2 Summary of all clinical decisions following US and MRE
results.
Clinical decision
Number of patients
Percentage detected in US
Percentage detected in MRE
Steroids
20
19 (95.0%)
18 (90.0%)
Antibiotic treatment
5
4 (80.0%)
4 (80.0%)
Surgery/drainage
6
6 (100%)
6 (100%)
Change of antibody
8
6 (75.0%)
8 (100%)
Other (e. g., 5-ASA*)
3
3 (100%)
3 (100%)
*5-ASA: 5-aminosalicylic acid.
Disease activity
37 of 42 patients (68.5%) with luminal bowel inflammation were correctly
detected by US. Five patients escaped detection with US, and in six cases, US
led to false-positive results. MRE identified 38 patients (67.9%) with
bowel inflammation correctly and missed four. There were also six false-positive
cases with MRE. Detailed information regarding ultrasound and MRE findings is
given in [Tables 3 ] and [4 ].
Table 3 Comparison of results of US and MRE in our study
population of 54 IBD patients, validated by the reference
standard.
Ultrasound
MRE
Total validated
tp*
tn
fp
fn
tp
tn
fp
fn
Small bowel and/or colon inflammation
37
6
6
5
38
6
6
4
42/54
(Neo) terminal Ileum inflammation
22
24
4
3
24
27
1
1
25/54
Colon inflammation
20
25
3
5
18
23
5
7
25/54
Stenoses
4
43
4
3
4
44
3
3
7/54
Fistulae
4
47
0
3
6
47
0
1
7/54
Conglomerate tumors
5
46
1
2
6
46
1
1
7/54
Abscesses
6
42
2
4
8
44
0
2
10/54
* tp: true positive; tn: true negative; fp: false positive; fn:
false negative.
Table 4 Statistical comparison of results of US and MRE
for detecting inflammation and complications in our IBD patient
cohort.
Sensitivity (%)
Specificity (%)
PPV (%)
NPV (%)
Accuracy (%)
Inflammation, overall
US
88.1
50.0
86.1
54.6
79.6
MRE
90.5
50.0
86.4
60.0
86.4
US+MRE
97.6
33.3
83.7
80.0
83.3
Inflammation, (neo) terminal Ileum
US
88.0
85.7
84.6
88.9
86.8
MRE
96.0
96.4
96.0
96.4
96.2
Inflammation, colon
US
80.0
89.3
87
83.3
84.9
MRE
72.0
82.1
78.3
76.7
77.3
Stenoses
US
57.1
91.5
50.0
93.5
87
MRE
57.1
93.6
57.1
93.6
88.9
Fistulae
US
57.1
100
100
95.0
94.4
MRE
85.7
100
100
97.9
98.2
Conglomerate
US
71.4
97.9
83.3
95.8
94.4
MRE
85.7
97.9
85.7
97.9
96.3
Abscesses
US
60.0
95.5
75.0
91.3
88.9
MRE
80.0
100
100
95.7
96.3
The findings correspond to a sensitivity of 88.1% (95% CI:
74.4–96.0%) and a specificity of 50.0% (95% CI:
21.1–78.9%) for US. MRE had a sensitivity of 90.5%
(95% CI: 77.4–97.3%) and specificity of 50.0%
(95% CI: 21.1% to 78.9%). The accuracy was 79.6%
(95% CI: 66.5% to 89.4%) for US and 81.5%
(95% CI: 68.6–90.8%) for MRE. There was no statistically
significant difference in sensitivities between US and MRE (p=1.0). When
including the discordant findings and thereby combining the results of US and
MRE, 41 of 42 patients with inflammation were detected and sensitivity increased
to 97.62% (95% CI: 87.43–99.94%). Specificity,
however, decreased to 33.3% (95% CI:
9.9–65.1%).
In the subgroup of patients who received colonoscopy, inflammation of the colon
or (neo-) terminal ileum was discovered in 18 of 25 patients. Sonography
detected 15 cases and MRE discovered 14. Using only histological inflammation as
a reference, the sensitivity for detecting inflammation of the colon or (neo-)
terminal was 83.3% (95% CI: 58.6 - 96.4%) and the
specificity was 28.6% (95% CI: 3.67–71.0%) for
US. MRE had a sensitivity of 77.8% (95% CI:
52.4–93.6%) and a specificity of 37.5% (95% CI:
8.5 to 75.5%). The accuracy was 68.0% (95% CI: 46.5 to
85.0%) for US and 65.4% (95% CI:
44.3–83.8%) for MRE. There was also no statistically significant
difference in sensitivities between US and MRE (p=1.0).
In our cohort, only 9 patients showed a
BMI≥25 kg/m2 . In this group, the
sensitivity was lower with 75.0% (95% CI:
34.9–96.8%) for US and 87.5% (95% CI:
47.4–99.7%) for MRE.
Disease extent
25 patients had confirmed disease involvement of the terminal or (neo-) terminal
ileum. US correctly detected 22 cases (88%) and missed 3 cases. In four
instances, US returned false-positive findings. MRE identified 24 cases
(96%) correctly. Only one patient with disease involvement of the
terminal ileum escaped detection with MRE. These findings correspond with a
sensitivity for detecting inflammation of the terminal or neo-terminal ileum of
88.0% (95% CI: 68.8–97.5%) and a specificity of
85.7% (95% CI: 67.3–96.0%) for US. The
sensitivity for MRE is 96.0% (95% CI:
79.7–99.9%) and the specificity is 96.4% (95%
CI: 81.7–99.9%). The accuracy is 86.8% (95% CI:
74.7–94.5%) for US and 96.2% (95% CI: 87.0 -
99.5%) for MRE. There was no statistically significant difference in
sensitivities between US and MRE (p=0.63).
25 patients had confirmed disease involvement of the colon. US detected 20
(80%) and missed 5 cases while 5 false-positive findings occurred. MRE
correctly identified 18 patients (72%) with colonic inflammation. Seven
patients were overlooked by MRE, and five findings were false positives. The
sensitivity for detecting disease involvement of the colon was 80.0%
(95% CI: 59.3–93.2%) for US and 72.0%
(95% CI: 50.6–87.9%) for MRE. The specificities were
89.3% (95% CI: 71.8–97.7%) and 82.1%
(95% CI: 63.1–93.9%), respectively. The accuracy was
84.9% (95% CI: 72.4–93.2%) for US and
77.3% (95% CI: 63.8–87.7%) for MRE. There was no
statistically significant difference in sensitivities between US and MRE
(p=0.73).
Complications
Complications defined as abscess, fistulae,
stenoses, or conglomerate tumors were present in 19 patients (35.2%). 9
patients (16.7%) had more than one complication.
Seven patients
had stenoses. US and MRE both correctly detected four of these case. In
addition, US returned four false-positive findings and MRE three. The
sensitivities for detecting bowel stenoses were 57.1 (95% CI:
18.4–90.1%) for both US and MRE with specificities of
91.5% (95% CI: 79.6–97.6%) for US and
93.6% (95% CI: 82.5–98.7%) for MRE. There was no
statistically significant difference in sensitivities between US and MRE for the
detection of stenoses (p=1.0).
Seven patients had one or more
fistulae. MRE correctly detected these in six cases and US in four. There were
no false-positive findings in US and MRE. The sensitivity was 57.1%
(95% CI: 18.4–90.1%) for US and 85.7%
(95% CI: 42.1–99.6%) for MRE. The specificity was
100% (95% CI: 92.4 to 100%) for both modalities. There
was no statistically significant difference in sensitivities between US and MRE
for the detection of fistulae (p=0.63).
There were seven patients
with conglomerate tumors. US identified five of these cases and MRE six. US and
MRE both led to one false-positive finding. The sensitivity was 71.4%
(95% CI: 29.0–96.3%) for US and 85.7%
(95% CI: 42.1–99.6%) for MRE. The specificity was
97.9% (95% CI: 88.7–100%) for US and MRE. There
was no statistically significant difference in sensitivities between US and MRE
for the detection of conglomerate tumors (p=1.0)
Abscesses were
found in ten patients, six identified via US and seven by MRE. In two cases, US
returned false-positive findings. The sensitivity was 60.0% (95%
CI: 26.2–87.8%) for US and 80.0% (95% CI:
44.4–97.5%) for MRE. The specificity was 95.5%
(95% CI: 84.5–99.4%) for US and 100%
(95% CI: 92.0–100%) for MRE. There was no statistically
significant difference in sensitivities between US and MRE for the detection of
abscesses (p=1.0).
Using multivariate logistic regression, known
markers for complications such as elevated CRP, WBC, or fever did not predict
the presence of abscesses or fistulae in patients with suspected inflammation on
sonography. Examples of matching US and MRE findings in four patients are shown
in [Figs. 1 ]
[2 ]
[3 ] for terminal ileitis ([Fig. 1a, b ]), left-sided colitis ([Fig. 2a–d ]) and terminal ileitis,
enteroenteric fistulae, and abscess ([Fig.
3a–c ]).
Fig. 1 4.9 year old patient with terminal ileitis: a )
longitudinal US image of terminal ileum with thickened wall, narrowed
lumen and hyperperfusionhypervascularity; b ) thickened bowel wall
with hyperperfusion of the terminal ileum (circle) and ascending colon
on postcontrast T1-weighted MRE images in coronal plane.
Fig. 2 5.9 year old patient with left-sided colitis: a )
longitudinal US image of descending colon with wall thickening
(0.58 cm) and total loss auf wall stratification (arrows);
b ) long-segment wall thickening and narrowed lumen of the
left colon (circle) on postcontrast T1-weighted MRE images in transverse
plane; c ) 56 year old patient with Crohn’s disease with
left-sided colitis: increase of vascularity; d ) 56 year old
patient with Crohn’s disease with left-sided colitis: hypoechoic
wall thickening > 4 mm.
Fig. 3 1.9 year old patient with terminal ileitis and
enteroenteric fistulae and abscess: a ) longitudinal US image of
terminal ileum with thickened wall and enteroenteric fistulae (arrow)
between the terminal ileum and another small bowel loop; b )
terminal ileum with wall thickening and hyperperfusion (circle) as well
as enteroenteric fistula (arrow) and large abscess (small arrow) on
postcontrast T1-weighted MRE images in transverse plane c )
transversal US image of large abscess with surrounding hyperechogenic
mesenterium and a small hypoechogenic area at 5 o’clock (arrow)
as a sign for developing fistula.
Discussion
This retrospective study compared the usefulness and accuracy of US and MRE for
detecting inflammation in IBD patients in a real-life setting.
In our study, bowel US and MRE revealed comparable sensitivity and specificity for
detecting intestinal inflammation in IBD patients with a disease flare. We observed
a trend towards higher sensitivity of MRE for detecting complications such as
stenoses and abscesses, although this finding did not prove to be statistically
significant. Most importantly, bowel US and MRE were shown to be equally important
and sufficient for making clinical decisions. Interestingly, when both methods are
combined, the sensitivity further increases to 97.6% while the specificity
decreases. Still, in certain indications, a combination of both methods might be
useful, e. g., when several disease locations are suspected.
When considering the most suitable method for each patient, clinicians have to factor
in several aspects and practical issues of different diagnostic tools. US is easy
to
use and can be repeated without limitations. Also, direct interaction with the
patient enhances the outcome of this method [14 ]. Our
study supports the fact that bowel US can achieve sufficient accuracy to diagnose
disease activity in IBD patients. On the contrary, bowel US has been shown to have
decreased sensitivity in adipose patients. This finding was confirmed in our study.
Additionally, follow-up data are needed to further evaluate the diagnostic accuracy
of US and MRE in detecting response to therapy. Only a few prospective studies to
assess therapy response via bowel US have been performed so far. They have shown
that despite reduced bowel wall thickness, a clear correlation between mucosal
healing defined via the gold standard and US findings could not be established [15 ]
[16 ].
Allocca et al. showed comparable accuracy for both methods when compared to the
standard reference method (colonoscopy) in guiding clinical decisions in patients
with Crohn´s disease [10 ]. However, another
recent study by Taylor et al. found higher sensitivity and specificity of MRE for
detecting small bowel disease in Crohn´s disease patients [17 ]. Most studies are focusing on Crohn´s
disease patients, but US is also recommended and used for detecting disease activity
and complications in patients with ulcerative colitis [18 ].
MRE is considered the current diagnostic standard for small bowel assessment and for
detecting complications in IBD patients [19 ]. MRE is
also suitable for therapy follow-up of IBD patients [8 ]. On the other hand, it entails higher invasiveness compared to bowel
US due to the need for the administration of intravenous contrast agent, and it is
more time-consuming. Our study indicates the superiority of MRE for finding
complications in IBD patients, although the difference was not statistically
significant. This suggests that MRE might be the preferred method in patients with
suspicion of complications.
The first and main limitation of our study is its retrospective nature. Second, we
only included a small number of patients with complications weakening the
statistical outcome in this patient group. Third, due to the retrospective analysis,
endoscopic data were only available for a limited number of patients. Due to a lack
of endoscopic findings, our reference standard is not as strong compared to other
prospective studies. However, we were able to rely on other clinical data and
markers to support the diagnosis of IBD flare.
MRE and bowel US are valid and useful diagnostic tools for diagnosing small and large
bowel disease in IBD patients. Each method is sufficient for making clinical
decisions, and both methods should not necessarily be combined since this leads to
less specificity. Bowel US is the appropriate tool for monitoring disease activity
in the small intestine and colon, whereas MRE should be employed in patients with
suspected complications.
