Crohn’s disease (CD) is a chronic inflammatory bowel disease (IBD) of rising incidence and prevalence [1
]. It has high complication rates [2
], requiring surgical treatment upon progression [3
] and resulting in a negative impact on patients’ quality of life [4
]. It can however be treated more successfully if the disease is detected early.
Superior soft tissue contrast resolution enables magnetic resonance imaging (MRI) to track bowel inflammation beyond the reach of the endoscope. However, magnetic resonance enterography (MRE) requires bowel distension by an oral hyperosmolar enteric contrast agent [5
], causing adverse effects sometimes leading to poor tolerance by patients [6
]. MRE is contraindicated in patients requiring general anesthesia. MRE in IBD requires the use of intravenous gadolinium contrast agents [5
], potentially causing systemic nephrogenic fibrosis [8
] and formation of gadolinium deposits in the brain and body tissues [9
]. This demonstrates that solutions allowing avoiding gadolinium administration are important.
Diffusion-weighted imaging (DWI) has shown a potential to replace contrast medium administration and to detect lesions before they come visible in conventional images [12
]. DWI images use diffusion gradient applied in three perpendicular axes, with several b values (for example, 0, 50, 800 s/mm2
]. Inflamed segments as high signal intensity (SI) zones are best identified in the DWI tracking images of high b value, most commonly, b = 800 s/mm2
]. Diffusion is measured quantitatively by the apparent diffusion coefficient (ADC). Nevertheless, in spite of the high sensitivity and specificity of adding DWI to the MR imaging protocol, the specificity of DWI alone in assessment of IBD is still low, being 39–61% [15
], since intact bowel walls also often present high SI in DWI tracking images of high b values (Figure 1
A derivation of DWI, diffusion-weighted imaging with background body signal suppression (DWIBS), was introduced for whole body imaging of patients in order to detect metastases and tumour relapse. DWIBS provides more uniform fat suppression through the use of short T1 inversion recovery (STIR) based approach. This is a free breathing technique permitting multiple number of signal acquisitions to average motion [17
]. Therefore, DWIBS provides a higher contrast-to-noise ratio, reduced image distortion, and better detection of subtle lesions [18
]. A disadvantage of DWIBS is poorer signal-to noise ratio (SNR) [19
], resulting in grainy image appearance. Despite the extensive use of DWIBS, there are very few studies regarding its use in assessment of the digestive tract. Tomizawa et al. performed studies on DWIBS in regards to assessing gall bladder walls and gastrointestinal tract [20
]. Several researchers included a free breathing DWI sequence in their MRE protocol for the assessment of inflammatory bowel disease [13
], however, no research data exist on comparison between DWI and DWIBS in bowel imaging. Whilst similar to other body tissue, the bowel wall presents a better resolution in DWIBS, compared to DWI sequence in both intact (Figure 1
) and inflamed (Figure 2
) bowel walls. Besides, DWIBS is more reproducible than DWI [26
], and ADC-DWIBS values are also unaltered by motion [27
]. Therefore, DWIBS could be beneficial over routinely used DWI in assessment of bowel inflammation.
The purpose of our study was to estimate the potential significance of using ADC measurements in DWI and DWIBS in patients, without preparing bowels with a hyperosmolar enteric contrast agent by setting the following tasks:
assessing ADC-DWI and ADC-DWIBS values in intestines without preparation (collapsed bowel) and after preparation with hyperosmolar enteric contrast agent (filled bowel);
assessing ADC-DWI and ADC-DWIBS values in the colon without preparation (in presence of intraluminal faeces) and after ingestion of enteric contrast agent (presence of it in the bowel lumen);
comparing the consistency between ADC values of DWI and DWIBS, in conditions with and without bowel preparation in both intestines and the colon, and analyzing the utilization of DWIBS in MR of bowel imaging.
2. Materials and Methods
2.1. Patient Population
This prospective study included 106 primary care patients (18–76 years old), referred to MRE from March 2015 until March 2018, due to dyspeptic complaints but with no clinical and/or morphological evidence of IBD. The inclusion criteria were: absence of typical IBD symptoms—diarrhea, bloody and/or mucous stool, severe and/or crampy abdominal pain and rectal involvement [28
The exclusion criteria were: age <18 years, fecal calprotectin (FC) level >200 μg/g, acute bowel infection, proven or previously diagnosed IBD, endoscopically proven enteropathy (e.g., coeliac disease, collagenous colitis etc.), present bowel tumor, and systemic diseases such as cystic fibrosis.
2.2. The Study
In this prospective observational cross-sectional study, ADC of DWI and DWIBS were assessed in bowel walls before and after preparation with hyperosmolar enteric contrast agent. Prior to bowel preparation, DWI and DWIBS scanning sequences were performed in all patients. Afterwards, patients were given enteric contrast agent, and scanned as per complete MRE protocol with DWI and DWIBS sequences included.
The study included two cohorts: (1) assessment of ADC-DWI and ADC-DWIBS intestinal walls before and after patient preparation, and (2) assessment of ADC-DWI and ADC-DWIBS in colonic walls before and after preparation. ADC measurements were only performed on bowel segments where high SI was present in DWI tracking images of b = 800 s/mm2. In order to compare ADC before and after bowel preparation, only patients with measurements both before and after preparation were included in the further data analysis. Similarly, for comparison between ADC-DWI and ADC-DWIBS, only patients with measurements performed in both DWI and DWIBS sequences were included in the further analysis. In both cohorts, data were grouped by the preparation state of the patient—non-prepared versus prepared bowels—and mutually compared.
All patients fasted for at least 6 h prior to MRE. After the initial scanning of DWI and DWIBS in prone position, patients were instructed to intake 1.250–1.500 l of 2.5% mannitol solution within 45–60 min, followed by full MRE exam in prone position. During the MRE exam and prior to DWI and DWIBS sequences, 20 mg dose of butylscopolamin was intravenously administered to reduce bowel peristalsis.
Patients were scanned with 1.5 T MRI system (Ingenia, Philips Medical Systems, Best, The Netherlands) using a 16-channel body coil. The applied DWI and DWIBS protocols were obtained from the Philips standard abdominal protocol and included in the protocol repository of the MRI system. To enable DWIBS-ADC measurements, the standard DWIBS protocol was amended by replacing a single b factor b = 1000 s/mm2 by three b factors 0 s/mm2, 600 s/mm2 and 800 s/mm2, consistent to DWI protocol.
The scanning parameters for DWI and DWIBS protocols are given in the Table 1
The study was approved by the Ethics committee of Riga Stradin’s University, and written informed consent was obtained from all patients. The permission number by the Ethics committee of Riga Stradin’s University is 6/10.09.2015.
2.3. 1st Cohort
The first cohort was formed of patients in whom high SI bowel walls in DWI tracking images of b = 800 s/mm2
were identified in at least one intestinal site. High SI regions in at least one intestinal region were identified in all 106 patients. Prior to bowel preparation, one collapsed jejunal segment in DWI and DWIBS image series was identified for each patient. After bowel preparation, one distended jejunal segment in DWI and DWIBS image series was identified for each patient. ADC values were measured in three sites per segment using 10–20 mm2
oval region of interest (ROI), both before and after preparation (Figure 3
2.4. 2nd Cohort
The second cohort was formed of patients in whom high SI bowel walls in DWI tracking images of b = 800 s/mm2
were identified in at least one colonic site. 78 of the 106 patients were identified to have high SI regions in at least one colonic region. Before bowel preparation in the DWI and DWIBS image series, one caecum or ascending colon segment with presence of intraluminal faeces was identified in each patient. After bowel preparation in the DWI and DWIBS image series, one caecum or ascending colon segment with presence of intraluminal mannitol was identified in each patient. ADC values were measured in three sites per segment using 10–20 mm2
ROI both before and after preparation (Figure 4
2.5. Statistical Analysis
Statistical analysis was performed using software Stata/IC (StataCorp LLC, Texas, USA), mean ADC values were compared with paired t-test, and 99% confidence intervals (CI) were calculated for differences. The statistical significance of differences between mean values within groups was determined using one-way ANOVA with Bonferroni correction. P value of <0.05 was considered to be statistically significant.
According to the current joint evidence-based guidelines by the European Chron’s and Colitis Organization as well as European Society of Gastrointestinal and Abdominal Radiologists, MRE imaging in IBD requires administration of contrast medium giving opportunity to estimate the bowel wall enhancement pattern [5
] as well as vasa recta engorgement commonly seen in CD [30
]. A significant advantage of using the gadolinium contrast agent is the ability to quantify Crohn’s disease activity [31
]. Nevertheless, according to literature data, DWI in detection of bowel inflammatory changes outperforms T1 dynamic series with intravenous gadolinium contrast agent [32
]. Therefore, DWI could be beneficial in case of diagnostic difficulties, for example, when, in Crohn’s disease, the inflamed bowel tissues are covered by intact mucosa [34
]. The drawback of DWI is its low specificity [15
In intact bowel walls, high intensity signal resembling inflammation in images of high b factors is a reason for the low specificity of DWI. This pattern is commonly explained with the T2 shine-through effect, where theoretically ADC should be high [35
]. Nevertheless, increased DWI signal along with low ADC is observed not only in inflamed but also in disease-free bowel walls [36
Upon reporting multiple MRE exams, we had several observations regarding the high SI bowel wall at the DWI tracking images of b = 800 s/mm2. Firstly, we noticed that intestinal SI was markedly higher in bowel wall before preparation, i.e., in totally collapsed bowel, compared to intestinal wall after preparation, i.e., in fully distended bowel. Secondly, in colon, SI was markedly higher in bowel wall after preparation, i.e., in presence of enteric contrast agent, compared to the colonic wall before preparation, i.e., in presence of faeces. In both situations, high SI bowel walls in the ADC map frequently presented low SI. These observations raised a question regarding ADC differences between bowel wall before and after patient preparation, in both intestines and colon.
Results from the 1st cohort comparing ADC of DWI and DWIBS between non-prepared (collapsed) and prepared (distended) intestines showed that ADC values in both DWI and DWIBS in the collapsed bowel sample were markedly lower than in distended bowel samples. As the bowel collapses, the number of cells per volume unit increases; however, the cells itself are not altered. Therefore, the volumes of intracellular and extracellular spaces were still constant, giving no reason for restricted diffusion. The measurement results could be explained by the partial volume effect. In prepared (filled) jejunal wall, the signal of very thin bowel wall was contaminated by the high intensity signal from the massive volume of enteric contrast agent, therefore, ADC value is high. However, in non-prepared (collapsed) bowel samples, the amount of high SI intraluminal content is less, therefore, the contamination of the intestinal wall signal is also less.
A similar explanation applies to the 2nd cohort comparing ADC of DWI and DWIBS between non-prepared (presence of low SI intraluminal faeces) and prepared (presence of mannitol) colon. The results showed that ADC values in both DWI and DWIBS were dependent on colonic intraluminal content and in the presence of low signal intensity faeces were nearly two times lower than in the presence of high signal intensity mannitol.
According to a number of studies, the performance of DWI varied among authors. The range of ADC values in normal bowel wall was 1.18–3.69 mm2
/s whereas in inflamed bowel segments—1.24–1.988 mm2
/s being significantly lower for 0.8–2.4 ×10−3
/s than in intact bowels. Several authors have also provided their cut-off ADC values for discriminating between inflamed and intact bowel walls. These values are mutually different and lie between ADC ranges of inflamed and intact bowel walls, except in one study where cut-off value lies within the range of the inflamed bowel. According to data from all researchers, ADC ranges of IBD and intact bowel do not mutually overlap [37
]. Most of these studies have been performed in prepared bowels. To the best of our knowledge, a team of Kiryu et al. is the only research group reporting ADC values in Crohn’s disease patients without patient preparation using free-breathing DWI (i.e., using STIR as fat suppression method). The reported ADC values show a similar trend, with that in prepared bowels being lower in disease-active segments and higher in disease-inactive areas (1.61 ± 0.44 × 10−3
/s versus 2.56 ± 0.51×10−3
/s in intestines, respectively) [13
]. This difference is high enough to concern the potential benefit of ADC measurements without bowel preparation. If the consistency of differences between ADC values of inflamed and normal bowels applies also in non-prepared bowel samples, this would allow a proper estimation of ADC in patients without preparation. Therefore, assessment of consistency between ADC values of non-prepared and prepared bowel walls was a goal of our research.
Comparing the obtained ADC values of bowel prior preparation and prepared bowel to literature data on ADC values of normal and inflamed bowels, it is obvious that, although the ADC values for the prepared bowels are markedly higher than ADC values in bowel prior to preparation, both ADC values of prepared and non-prepared bowel loops overlap with the ADC range of inflamed bowel provided in literature [37
]. Thus, the applicability of ADC in non-prepared bowel samples is still questionable and might be related to scanning conditions and parameters in our institution. to fully compare the extent of overlap in ADC ranges for non-prepared bowel and inflamed bowels, the ADC values for CD have to be obtained in our institution under identical conditions.
A questionable issue is how much mannitol itself changes ADC, by impacting cells due to concentration gradient between bowel epithelium cells of lower and higher osmolarity substance, and the expelling fluid from cells resulting in a lowered packed cell volume [38
] and osmotic diarrhoea [6
]. Nevertheless, we assume that as the bowel wall filled with mannitol is a very thin (<3 mm), the contribution to signal intensity by osmotic influence of mannitol is negligible.
By mutually comparing ADC-DWI and ADC-DWIBS, we observe no statistically significant difference in the wall of prepared bowels, both regarding intestines and the colon. On the contrary, ADC values in non-prepared bowels and both intestines and colon, is markedly lower than in the prepared bowel samples. In the prepared bowel, the wall signal was influenced with a high large amount of fluid. In the non-prepared bowel, there was a presence of high-viscosity intraluminal content—chime in intestines and faeces in colon. DWI and DWIBS sequences differed when fat suppression techniques were used. In DWI, CHESS (CHEmically Selective Saturation) or SPIR sequences suppress fat selectively. The STIR technique used in DWIBS was based on the T1 relaxation time of tissues [39
], and suppressed signals from all substances of short T1 values such as proteinaceous, viscous and mucous substances, including chime and faeces, thus being non-selective [40
]. Therefore, ADC-DWIBS values are lower in the presence of the bowel content, comparing to ADC-DWI, and the probability that the ADC-DWIBS range will overlap the ADC range of inflamed bowel is higher. Therefore, based on the findings of this study, we still recommend preferring SPIR-based ADC measurements and using ADC values of STIR-based ADC in non-prepared bowel with caution.
Our study had several limitations. 1) Measurements were performed by one radiologist not assessing inter-observer agreement, and in such a small volume ADC values are reported to be hardly reproducible [37
] as they rest on subjectivity. Nevertheless, data from ADC measurements in liver imaging suggested better reproducibility of free-breathing DWIBS over respiratory-triggered DWI [26
], which could be also proven better for bowel walls, but requires further investigation. 2) Typically, achievable DWI resolution is on the order of 2 × 2 × 2 mm3
]. The pixel size used in our standard protocols (see Table 1
) could be large for tiny structures, such as the bowel wall. ADC values are therefore markedly impacted by the partial volume effect being very approximate and can be used only for reference but not as absolute values. 3) In intestines, the most uniform luminal distension was present in jejunum, which was therefore chosen for intestinal measurements, however, ileum is the main location of CD. The terminal loop of ileum also has different morphological patterns—abundance of lymphoid tissues [43
which also could influence ADC measurements. In the colon, measurements were performed only in walls of the caecum and the ascending colon, since presence or mannitol was mainly observed in these locations. 4) Location of the sites with high SI signal in DWI tracking images of b = 800 s/mm2
was not consistent among the series, therefore, measurements could not be performed precisely at the same locations. 5) We did not pay special attention to the T2 shine through effect of bowel walls, and measured ADC values in DWI tracking images of b = 800 s/mm2
regardless of signal appearance in ADC map. 6) Our goal was to observe properties of DWI-ADC and DWIBS-ADC in sites of bowel walls showing high SI at DWI tracking images of b = 800 s/mm2
resembling bowel inflammation, whereas we did not consider other signs of bowel inflammation like oedema, increased bowel wall thickness, contrast enhancement, etc., which of course were absent in patients with no presence of IBD as required by the study.