To investigate the application of free-breathing diffusion-weighted MR imaging (DWI) to the assessment of disease activity in Crohn's disease.
To investigate the application of free-breathing diffusion-weighted MR imaging (DWI) to the assessment of disease activity in Crohn's disease.
Thirty-one patients with Crohn's disease were investigated using free-breathing DWI without special patient preparation or IV or intraluminal contrast agent. The bowel was divided into seven segments, and disease activity was assessed visually on DWI. For quantitative analysis, the apparent diffusion coefficient (ADC) was measured in each segment. The findings of a conventional barium study or surgery were regarded as the gold standard for evaluating the diagnostic ability of DWI to assess disease activity.
Upon visual assessment, the sensitivity, specificity, and accuracy for the detection of disease-active segments were 86.0, 81.4, and 82.4%, respectively. In the quantitative assessment, the ADC value in the disease-active area was lower than that in disease-inactive area in small and large bowels (1.61 ± 0.44×10−3 mm2/s versus 2.56 ± 0.51 × 10−3 mm2/s in small bowel and 1.52 ± 0.43 × 10−3 mm2/s versus 2.31 ± 0.59 × 10−3 mm2/s in large bowel, respectively, P<0.001).
Free-breathing DWI is useful in the assessment of Crohn's disease. The accuracy of DWI is high in evaluating disease activity, especially in the small bowel, and the ADC may facilitate quantitative analysis of disease activity. J. Magn. Reson. Imaging 2009;29:880–886. © 2009 Wiley-Liss, Inc.
CROHN'S DISEASE is an inflammatory disorder of unknown cause that affects mainly young people. It may involve any part of the gastrointestinal tract, and the course is chronic, relapsing, and remitting. Long-standing inflammation may cause several complications in the bowel, including obstruction, stricture, fistula, and abscess. In addition, increased risk for malignancy at the area of inflammation has been reported in the small and large bowel (1, 2). The assessment of inflammatory activity plays a crucial role in the management of patients with Crohn's disease.
Recently, immunomodulation by antibodies to tumor necrosis factor α has been successfully used to treat patients with severe active inflammation (3, 4). This treatment may cause substantial adverse effects (5) and, thus, should be applied exclusively to patients with active disease. Therefore, the need for repeated evaluation to monitor disease activity is increasing, making safe examination with minimal burden to the patient desirable for the assessment of disease activity of Crohn's disease. Barium enteroclysis remains the gold standard in the work-up of inflammatory disease of the small bowel, because endoscopic examination of its entire length is difficult (6); however, this procedure is usually stressful to the patient, and the radiation exposure cannot be ignored, especially in young people, whom Crohn's disease often affects. Moreover, the examination is operator-dependent.
To assess the activity of Crohn's disease, imaging studies, such as leukocyte scintigraphy (7), computed tomography (8, 9), MRI (10–16), sonography (17), and 18F-fluorodeoxyglucose positron emission tomography (18), have been performed. Among these imaging modalities, except for sonography and MRI, radiation exposure is a serious concern, as well as conventional barium study. With its high intrinsic contrast resolution, MRI is regarded as a promising tool for the detection of intestinal abnormalities. MRI is sensitive for detecting soft-tissue inflammation, and several groups have reported the usefulness of T1-weighted gadolinium-enhanced MRI in the assessment of disease activity (12–15). Increased capillary permeability in the inflamed bowel wall is assumed to increase the contrast enhancement of intravenously administered gadolinium contrast material. However, gadolinium contrast materials can cause severe adverse effects, including nephrogenic systemic fibrosis (19, 20). MR enteroclysis imaging is also used in the assessment of Crohn's disease; however, it requires special preparation to distend the small bowel, using intubation of the duodenum or proximal small bowel and oral administration of a contrast agent (11). Although MR enterography obviates the need for nasoenteric intubation, it requires the intake of a large amount of fluid (1.5–2 L) (10).
Diffusion-weighted MR imaging (DWI) reflects molecular diffusion, which is the thermally induced Brownian motion of water molecules, without the administration of contrast material. Changes in proton self-diffusion are an early indicator of alterations in cellular homeostasis in acute ischemic stroke (21), and DWI has been established as an important tool in the detection of early ischemic changes in clinical settings. DWI also has the capability to detect inflammatory foci (22–24). In body imaging, DWI is used in solid organs such as liver, pancreas, spleen, and kidney (25–28). The quantitative analysis using apparent diffusion coefficient (ADC) also has been assessed in the liver (29, 30). In gastrointestinal tract imaging, DWI has not been commonly used because the image quality is degraded severely by bulk motion artifacts related to intestinal motion. Recently, free-breathing DWI has been applied to whole-body evaluation (31). In this technique, free-breathing affords multiple excitations in data acquisition and signal averaging of reconstructed images, which requires an extended acquisition period. A short tau inversion recovery (STIR) sequence is used for the purpose of fat suppression.
We hypothesized that an actively inflamed lesion in the bowel could be detected by using free-breathing DWI. Owing to the lack of ionizing exposure and contrast administration, this technique may be favorable for assessing patients with Crohn's disease. To test our hypothesis, we compared DWI findings with those of conventional barium examination or surgery as a gold standard in Crohn's disease. The purpose of this study was to evaluate the usefulness of DWI in the assessment of Crohn's disease activity.
From September 2005 to April 2006, consecutive patients with a proven diagnosis of Crohn's disease and suspected active inflammation were referred for MRI by gastroenterologists. The final diagnosis was established based on conventional barium examination of the small and large bowels, or surgery performed within 1 month before or after MRI. After patients lacking a diagnostic standard were excluded, a total of 31 patients (25 male, 6 female) with a mean age of 37.7 years (range, 24–65 years) were included in the study. The final diagnosis was based on surgical findings in 14 patients and conventional barium study in the rest of the 17 patients. Among the 31 patients, 18 had a history of intestinal surgery: ileectomy in 2 patients, terminal ileectomy in 8 patients, terminal ileectomy and colectomy in 7 patients (2, ascending colon; 1, transverse colon; 3, descending colon; and 1, sigmoid colon), and terminal ileectomy and coloproctectomy in 1 patient.
MRI was performed using a 1.0 Tesla (T) MR scanner (Siemens, Erlangen, Germany) equipped with a phased-array coil. Patients were required to fast for 2 h, but there was no other special patient preparation for MRI. First, transverse and coronal T2-weighted half-Fourier single-shot turbo spin-echo (HASTE) images were acquired during a breath-hold, with the following parameters: repetition time/echo time (TR/TE), 1000/90–100 ms; slice thickness/slice interval, 8–10/0.8–1.0 mm in transverse and 5/0.5 mm in coronal view; field of view, 32–35 cm; matrix, 256 × 210–256; and bandwidth, 180 Hz/pixel. In transverse imaging, 25–30 sections were acquired during a breath-hold of 25–30 s. In coronal imaging, 20–25 sections were acquired during a breath-hold of 20–25 s. Subsequently, free-breathing DWI based on a previous study by Takahara et al (31) was performed. Transverse images were acquired during free-breathing by using STIR and echo planar imaging with the following parameters: TR/TE, 7700/73 ms; inversion time, 160 ms; slice thickness/slice interval, 8–10/0.8–1.0 mm; field of view, 36 cm; matrix, 128 × 88; number of excitations, 4; bandwidth, 1345 Hz/pixel. A diffusion gradient was applied in three perpendicular axes. A total of 30 sections were acquired during 6 min 11 s. A set of images with different b-values (0, 50, and 800 s/mm2) was acquired, and ADC maps were generated. With the limitation of the craniocaudal length of the phased array coil, the imaging range always covered the pelvis but not the entire abdomen. Typically, the upper level of the imaging range was near the upper pole of the kidney.
The final diagnosis was based on conventional barium study in 17 patients. A barium study of the small bowel was performed with enteroclysis (32), using a nasojejunal tube and methylcellulose and barium sulfate as contrast agents. A barium enema study of the large bowel was performed using a double-contrast technique. Both procedures were undertaken by one of three operators, each with more than 10 years of experience. All operators were blinded to the MRI findings. The radiographs from the barium enteroclysis and enema were evaluated by two experienced reviewers in consensus. According to the classification of Maglinte et al (33), intermediate (nodular pattern and ulcerations) and advanced (ulcer–nodular pattern) lesions were regarded as active disease.
A total of 14 patients underwent surgical exploration. Surgery was performed by experienced gastrointestinal surgeons for inflammatory bowel disease at the coloproctology center of the host hospital. Surgeons inspected carefully all segments during the procedure, and portions of the bowel suspected of having inflammatory activity were resected. The disease activity was confirmed by pathological examination. Acute inflammatory infiltrates observable upon pathological examination were regarded as proof of active lesions.
Visual assessment was performed by two radiologists blinded to patient information in consensus, under the direction of another radiologist. All images were reviewed on a picture archiving and communication system (PACS; eFilm Medical, Toronto, Ontario, Canada) using two 1200 × 1600 color LCD monitors (TOTOKU Electric Co, Nagano, Japan) and patient information was removed from the monitor during the visual assessment. The study conductor adjusted the initial window setting for each patient, and the readers altered the setting at their discretion during the review. There was no communication between the study conductor and the reviewers during the visual assessment. The bowel was divided into seven segments: jejunum, ileum, ascending colon, transverse colon, descending colon, sigmoid colon, and rectum. Jejunal and ileal loops were distinguished by their different locations and different folding patterns on T2-weighted HASTE images. The signal intensity of each segment on DWI with b-value of 800 s/mm2 was graded as 1, 2, or 3 for intensity as low as, slightly higher than, or definitely higher than the adjacent normal-appearing loops, respectively. In segments with nonhomogeneous intensity, the reviewers determined the grade based on the brightest area. To understand the grading system, the two reviewers were trained on typical cases prepared by the study conductor, with a 2-wk interval between the training session and the reading session. Reviewers conferred to reach consensus. To avoid contamination of T2 shine-through, DWI with b-values 0 and 50 s/mm2 were also used in the visual assessment. The location of inflamed lesion on barium study was translated on DWI images by interactive viewing of transverse and coronal T2-weighted HASTE images on PACS.
Segments graded as 2 or 3 were considered to have active disease, and segments graded as 1 were considered inactive. The sensitivity, specificity, and accuracy of MRI assessment of disease-active segments were determined using the findings of surgery or conventional barium examination as a gold standard. For patients who underwent surgery, surgical findings complemented by pathological findings were regarded as the standard; the findings of conventional barium examination were regarded as the standard for the remaining patients. In addition, complications such as fistula and abscess were recorded when noted in the visual assessment of MRI.
For quantitative assessment, the ADC was calculated after the reading session. A region of interest (ROI) of ∼1 cm2 in the area of highest signal intensity was drawn on DWI of each segment, and the ROI was copied to the same location on ADC map; thus, the mean ADC in the ROI was defined as the ADC for the segment. The evident area of intestinal air on DWI or T2-weighted HASTE images was avoided for measuring the ADC. The measurement of the ADC was performed by the study conductor prospectively, who was blinded to the results of the prior study. The ADC of complication was also measured. The Mann–Whitney U test was used to statistically compare the ADC between active and inactive segments. Statistical analyses were carried out using commercially available software (JMP™, SAS Institute, Cary, NC). All values are expressed as means ± standard deviation (SD).
In the current study, 31 segments of jejunum, 29 of ileum, 23 of ascending colon, 20 of transverse colon, 26 of descending colon, 29 of sigmoid colon, and 30 of rectum were evaluated. Five segments of transverse colon were located outside the of the DWI scanning range, which was limited by the craniocaudal length of the phased array coil, and were not assessed. Among the evaluated segments, 43 (22 in the small bowel and 21 in the large bowel) were judged as having active inflammation, based on the findings of a conventional barium study or pathological examination (24 by conventional barium and 19 by pathological examination). There were no complications during the performance of the barium examinations. In all patients, free-breathing DWI provided images with a quality sufficient for visual evaluation, and 124, 35, and 29 segments were graded as 1, 2, and 3, respectively (Fig. 1).
The sensitivity, specificity, and accuracy of the MRI assessment of disease-active segments on DWI are shown in Table 1. Overall, the sensitivity, specificity, and accuracy were 86.0, 81.4, and 82.4%, respectively. The sensitivity, specificity, and accuracy were 86.4, 97.4, and 93.3%, respectively, in the small bowel, and 85.7, 75.7, and 77.3%, respectively, in the large bowel. The small bowel showed better specificity and accuracy than the large bowel. The accuracy was 100% in the jejunum, 86.2% in the ileum, 82.6% in the ascending colon, 85.0% in the transverse colon, 80.8% in the descending colon, 72.4% in the sigmoid colon, and 70.0% in the rectum. Thus, the accuracy was lower in the lower segment than in the upper segment. False-positive findings, that is, abnormally high DWI signal areas in inactive segments, tended to be more frequent in the lower segment than in the upper segment. There were no disease-active cases in the rectum; therefore, sensitivity was not available. Fistulas in five and abscesses in two patients were observed as increased signal areas on DWI (Fig. 2).
|Location||Sensitivity, %||Specificity, %||Accuracy, %|
|Small bowel||86.4 (19/22)||97.4 (37/38)||93.3 (56/60)|
|Jejunum (n = 31)||100 (5/5)||100 (26/26)||100 (31/31)|
|Ileum (n = 29)||82.4 (14/17)||91.7 (11/12)||86.2 (25/29)|
|Large bowel||85.7 (18/21)||75.7 (81/107)||77.3 (99/128)|
|Ascending colon (n = 23)||100 (5/5)||77.8 (14/18)||82.6 (19/23)|
|Transverse colon (n = 20)||100 (2/2)||83.3 (15/18)||85.0 (17/20)|
|Descending colon (n = 26)||66.7 (4/6)||85.0 (17/20)||80.8 (21/26)|
|Sigmoid colon (n = 29)||87.5 (7/8)||66.7 (14/21)||72.4 (21/29)|
|Rectum (n = 30)||N/Aa (0/0)||70.0 (21/30)||70.0 (21/30)|
|Total||86.0 (37/43)||81.4 (118/145)||82.4 (155/188)|
The ADCs are shown in Table 2. Overall, the ADC in the disease-active segments was significantly lower than that in the disease-inactive segments (1.57 × 10−3 ± 0.44 × 10−3 mm2/s versus 2.38 × 10−3 ± 0.58 × 10−3 mm2/s, respectively; P < 0.0001). The ADC was also lower in the disease-active segments than in the disease-inactive segments for the small and large bowels analyzed separately (small bowel: 1.61 × 10−3 ± 0.44 × 10−3 mm2/s versus 2.56 × 10−3 ± 0.51 × 10−3 mm2/s, respectively; large bowel: 1.52 × 10−3 ± 0.43 × 10−3 mm2/s versus 2.31 × 10−3 ± 0.59 × 10−3 mm2/s, respectively; P < 0.001). For each segment type, the ADC was lower in segments showing active disease than in those showing inactivity, although the difference did not reach statistical significance for the transverse colon. The ADC in the disease-inactive segments in the large bowel was significantly lower than that in the small bowel (2.31 × 10−3 ± 0.59 × 10−3 mm2/s versus 2.56 × 10−3 ± 0.51 × 10−3 mm2/s, respectively; P = 0.011). The ADC values in abscesses in two patients were 1.60 × 10−3 mm2/s and 1.68 × 10−3 mm2/s. Because the fistula was a narrow canal structure on DWI, drawing the ROI of ∼1 cm2 on the fistula was not possible. On the ADC map, the fistula was obscure.
|Location||ADC value (mm2/s)||Accuracy|
|Small bowel||1.61 ± 0.44||2.56 ± 0.51||<0.001|
|Jejunum (n = 31)||1.32 ± 0.40||2.54 ± 0.47||0.0007|
|Ileum (n = 29)||1.69 ± 0.43||2.61 ± 0.62||0.0008|
|Large bowel||1.52 ± 0.43||2.31 ± 0.59||<0.001|
|Ascending colon (n = 23)||1.32 ± 0.21||2.35 ± 0.65||0.0017|
|Transverse colon (n = 20)||1.66 ± 0.34||2.44 ± 0.62||0.1015|
|Descending colon (n = 26)||1.59 ± 0.52||2.14 ± 0.41||0.0191|
|Sigmoid colon (n = 29)||1.57 ± 0.56||2.24 ± 0.68||0.0218|
|Rectum (n = 30)||N/A||2.37 ± 0.60||N/A|
|Total||1.57 ± 0.44||2.38 ± 0.58||<0.0001|
In this study, we demonstrated the usefulness of free-breathing DWI in the assessment of active Crohn's disease. The bowel segments with active inflammation were detected as those containing high signal intensity on DWI. Based on visual evaluation, the overall accuracy was 82.4%, with an accuracy of 93.3% in the small bowel, where the assessment of disease activity is difficult because of poor accessibility. Previous MRI studies have also demonstrated high sensitivity in assessing the activity of Crohn's disease in the small bowel (15, 16); however, contrast materials were used in those studies. In the current study, neither special patient preparation nor oral/intravenous administration of contrast materials was required. The course of Crohn's disease is chronic, relapsing, and remitting, therefore, reducing the patient burden is particularly important. Compared with earlier MRI methods, the method using DWI described in the present study appears to be less burdensome.
Quantitative assessment using the ADC is also feasible with the DWI technique. In the small and large bowels, the ADC in the disease-active segments was lower than that in the disease-inactive segments. In clinical settings, a scoring system based on clinical and laboratory parameters is used to evaluate disease activity in Crohn's disease. However, that system reflects the degree of global inflammation in each patient, and regional assessment is not feasible for multifocal intestinal lesions, which are common in Crohn's disease. Moreover, the classification of disease activity based solely on clinical and laboratory parameters is not sufficiently reproducible (34). Visual assessment with DWI allows the determination of regional disease activity, and the ADC may quantitatively support the visual assessment. Quantitative analysis may aid the monitoring of disease conditions and therapeutic effects, and the usefulness of the ADC in long-term follow-up deserves further investigation.
The accuracy of DWI was lower in the large bowel (77.3%) compared with the small bowel. Although the sensitivity was not disappointing in the large bowel (85.7%), the specificity was relatively low (75.7%). In the large bowel, segments without active inflammation showed high intensity on DWI in some cases, and this appeared to decrease the specificity. In parallel with the results of the visual assessment, the ADC of the disease-inactive segments in the large bowel was significantly lower than that in the small bowel.
The increased DWI signal intensity and decreased ADC in disease-inactive segments may be partly explained by the high viscosity of the intestinal contents. The brain abscess shows decreased ADC and the pus consistency itself could account for the restricted diffusion (22). A similar mechanism may contribute to the observed result. In the liver DWI assessment, the lower liver ADC was found in patients with established fibrosis and inflammation due to chronic liver disease when compared with normal controls (29). Although fibrosis is not predominant in the active phase of Crohn's disease, a similar phenomenon may occur in the active lesion in Crohn's disease. The reason underling the observed lower ADC in large bowel than in small bowel is not unclear; however, it may be due to the higher viscosity in large bowel than in small bowel. Based on our data, the need for investigation by other modalities is greater in the large bowel than in the small bowel; in addition, access is much easier for the large bowel than for the small bowel. The high diagnostic accuracy of DWI in assessing lesions of the small bowel should be beneficial in managing patients with Crohn's disease.
In this study, free-breathing DWI was used to assess the Crohn's disease. In body DWI, other DWI techniques such as breath-hold and respiratory-triggered scanning, which can decrease the effect of respiratory motion, has been also applied (30, 35, 36) and may be an alternative method in the assessment of Crohn's disease. In the assessment using quantitative analysis, such as ADC measurement, reproducibility is important. Kwee et al evaluated the reproducibility of ADC measurements in liver imaging and concluded that respiratory-triggered DWI shows less reproducibility than breath-hold and free-breathing DWI (37). To adopt their result to the assessment of gastrointestinal DWI requires further evaluation, because the imaging area is not identical and peristalsis is a very concerning issue in gastrointestinal imaging. In addition to the reproducibility of imaging technique, there are other issues to be assessed for DWI in the gastrointestinal tract, such as optimization of b-values, evaluation of need for antiperistaltic agents or intestinal cleansing, and so on. Further study is required to establish the optimal factors of DWI in the gastrointestinal tract.
In the current study, complications such as fistulas and abscesses were detected by DWI. They showed high signal intensity against a suppressed background signal on DWI, clearly delineating these complications. Because these lesions frequently are poorly visualized on conventional barium studies, DWI may play an important role in detecting fistulas and abscesses as well as in evaluating active Crohn's disease. Although the fistula was delineated clearly on DWI, it was obscure on the ADC map. This finding may be due to the close ADC value of a fistula to that of a surrounding noninflamed structure.
The current study has a few limitations. The number of patients in the study was relatively small, and we only studied symptomatic individuals. We did not acquire DWI data from Crohn's disease patients who had active inflammatory lesions but who did not exhibit obvious symptoms. The visual assessment was performed by the consensus of the reviewers, and interobserver agreement between independent readings was not evaluated. Very few studies deal with DWI of the gastrointestinal tract, and we decided to carry out visual assessment in consensus. To determine the reproducibility of the assessment of DWI in Crohn's disease, the assessment of interobserver agreement is needed in future studies. The imaging spatial resolution was limited for DWI in the current study (128 in-plane matrix); therefore, small inflamed foci may have been obscured on DWI.
In conclusion, free-breathing DWI allows the disease activity in Crohn's disease to be assessed without the use of contrast media. Visual assessment of DWI provides high accuracy, especially in the small bowel, and the calculation of the ADC may facilitate the quantitative analysis of disease activity. Considering its relatively light patient burden, free-breathing DWI may contribute to the long-term management of patients with Crohn's disease.