Gastrointestinal Imaging Russell N. Low, MD Christopher P. Sebrechts, MD Douglas A. Politoske, MD Michael T. Bennett, MD Sergio Flores, MD Richard J. Snyder, MD Jeffrey H. Pressman, MD

Index terms: Crohn disease, 70.262 Magnetic resonance (MR), comparative studies, 70.121411, 70.121412, 70.121415, 70.121416, 70.12143 Published online before print 10.1148/radiol.2223010811 Radiology 2002; 222:652– 660 Abbreviations: GRE ⫽ gradient echo RARE ⫽ rapid acquisition with relaxation enhancement SE ⫽ spin echo 1

From the Sharp and Children’s MRI Center (R.N.L., C.P.S.), and Departments of Radiology (R.N.L., C.P.S.) and Gastroenterology (D.A.P., M.T.B., S.F., R.J.S., J.H.P.), Sharp Memorial Hospital, 7901 Frost St, San Diego, CA 92123. Received April 19, 2001; revision requested June 5; revision received July 31; accepted September 17. Address correspondence to R.N.L. (e-mail: [email protected]).

©

RSNA, 2002

Crohn Disease with Endoscopic Correlation: Single-Shot Fast Spin-Echo and Gadoliniumenhanced Fat-suppressed Spoiled Gradient-Echo MR Imaging1 PURPOSE: To compare T2-weighted breath-hold single-shot fast spin-echo (SE) and gadolinium-enhanced spoiled gradient-echo (GRE) MR imaging with contrast material administered orally and rectally for evaluating patients with Crohn disease. MATERIALS AND METHODS: Twenty-eight patients with Crohn disease received 2% barium sulfate and water enema. The abdomen and pelvis were imaged with transverse and coronal single-shot fast SE and gadolinium-enhanced spoiled GRE MR imaging. Two radiologists reviewed the two types of images for bowel disease. The extent, severity, and conspicuity of the disease were determined. Proof of bowel disease at MR imaging was compared with that at endoscopy, barium study, and surgery. Statistical analysis was performed with the McNemar test. RESULTS: Twenty-five of 28 patients had proven abnormal bowel segments. The per-patient sensitivity of gadolinium-enhanced spoiled GRE MR imaging for the two radiologists was 100% and 96% versus 60% and 60% (P ⬍ .05) with single-shot fast SE MR imaging. Gadolinium-enhanced spoiled GRE MR images depicted more segments (54 and 52 of 61 segments; sensitivity, 89% and 85%, respectively) of the diseased bowel than did single-shot fast SE MR images (31 and 32 of 61 segments; sensitivity, 51% and 52%, respectively; P ⬍ .001). Severity of Crohn disease was correctly depicted at gadolinium-enhanced spoiled GRE imaging in 93% of patients versus in 43% of patients at single-shot fast SE imaging. CONCLUSION: In patients with Crohn disease, gadolinium-enhanced fat-suppressed spoiled GRE MR imaging better depicted the extent and severity of intestinal disease compared with single-shot fast SE imaging. ©

Author contributions: Guarantor of integrity of entire study, R.N.L.; study concepts, R.N.L., C.P.S.; study design, R.N.L.; literature research, R.N.L.; clinical studies, J.H.P., D.A.P., M.T.B., S.F., R.J.S.; data acquisition and analysis/interpretation, R.N.L., C.P.S.; statistical analysis, R.N.L.; manuscript preparation and definition of intellectual content, R.N.L.; manuscript editing, R.N.L., C.P.S.; manuscript revision/review, all authors; manuscript final version approval, R.N.L., C.P.S.

652

RSNA, 2002

Crohn disease is a chronic inflammatory disease of the gastrointestinal tract of unknown cause characterized by aphthous ulceration, cobblestoning, strictures, and fistula formation (1,2). Changes in the bowel wall are typically discontinuous and asymmetric and can be depicted on cross-sectional imaging studies, which can be used to supplement information obtained from endoscopic and barium studies. Helical computed tomography (CT) continues to be used as the predominant cross-sectional imaging examination in patients with Crohn disease due to its proven efficacy in depicting intestinal disease and extraintestinal complications (3,4). Magnetic resonance (MR) imaging of Crohn disease has been reported by many authors (5–11) because of the inherently high contrast resolution in the depiction of inflammatory changes of Crohn disease. Fast MR pulse sequences that facilitate T1- and T2-weighted breath-hold imaging are now routinely used for imaging of the abdomen and pelvis and

have shown promise in the evaluation of the gastrointestinal tract (17–21). Acquisition of these images during suspended respiration results in improved image quality, with a reduction in artifacts from physiologic motion. Half-Fourier rapid acquisition with relaxation enhancement (RARE) MR imaging provides breathing-independent T2-weighted images with excellent depiction of abdominal and pelvic anatomy. These rapidly acquired T2-weighted images have been proposed for evaluation of the gastrointestinal tract (17–21). Gadolinium enhancement of the inflamed bowel segments involved with Crohn disease has been described (6 –11) as a sensitive means of depicting mural changes. Kettritz et al (6) showed that the clinical severity of Crohn disease correlated with the product of the degree of contrast enhancement with use of an intravenously administered gadoliniumbased contrast agent, wall thickness, and length of the diseased bowel segment. Negative intraluminal contrast agents such as water or iron oxides combined with intravenous gadolinium-based contrast agents accentuate mural enhancement by distending bowel loops with a hypointense contrast material (12–16). The combination of dilute oral barium sulfate and an intravenous gadoliniumbased contrast agent has been previously described for MR imaging of benign and malignant gastrointestinal tract diseases (8,9). The purpose of this study was to compare T2-weighted breath-hold single-shot fast spin-echo (SE) and gadolinium-enhanced spoiled gradient-echo (GRE) MR imaging with oral and rectal contrast material for evaluating patients with Crohn disease.

MATERIALS AND METHODS Patients Fifty-six consecutive patients with known or suspected Crohn disease were referred for MR evaluation between February 1999 and April 2001. All imaging studies were performed at Sharp and Children’s MRI Center. All patients were experiencing symptoms of Crohn disease at the time of the imaging studies. Twenty patients with subsequent clinical diagnosis of infectious ileocolitis (n ⫽ 4), nonspecific colitis (n ⫽ 7), right lower quadrant adhesions (n ⫽ 1), and irritable bowel syndrome (n ⫽ 8) were excluded from the study group. Eight patients with clinical Crohn disease were also excluded Volume 222

䡠

Number 3

owing to lack of proof of disease. The remaining 28 patients (18 women, 10 men; mean age, 39 years; age range, 23–58 years) form the study group. All MR examinations were ordered by the patient’s gastroenterologist as part of the clinical evaluation to help determine the extent of disease and to depict extraintestinal complications of Crohn disease. MR images and medical records were reviewed (R.N.L., C.P.S.) in a retrospective manner. A summary of the conclusions of this study will be submitted to our hospital as a clinical quality improvement project. Informed consent and approval were not required by our institution’s investigational review board.

Proof of Disease All patients had histopathologic proof of Crohn disease. We performed correlation between clinical data and multiple examinations, including colonoscopy, biopsy, or both (28 patients), upper gastrointestinal endoscopy, biopsy, or both (eight patients), upper gastrointestinal and small bowel examination (17 patients), and surgical exploration (six patients). Proof of disease was obtained within 4 weeks of MR imaging. Endoscopic proof was obtained after MR examination in 12 patients and prior to MR examination in 16 patients.

MR Imaging MR imaging was performed with a 1.5-T imager (Signa; GE Medical Systems, Milwaukee, Wis) with high-performance gradients (23 mT/m, 120 mT/m/sec) equipped with a software package (version 8.2.5– 8.3.1). Body coil imaging was used in all patients. Patients fasted for 3 hours prior to MR examination. Starting 45 minutes prior to imaging, patients were given 1,350 mL of 2% dilute barium sulfate (Readi-Cat 2; E-Z-Em, Westbury, NY) orally and were instructed to drink one bottle (450 mL) of dilute barium every 15 minutes. Later in the study, patients were instructed to drink 900 mL of dilute barium for 20 –30 minutes. Water (500 –1,000 mL) was administered rectally for colonic distention just prior to imaging. Patients were given an additional 200 mL of water orally to distend the stomach, and 1 mg of intravenous glucagon (Glucagon for Injection; Eli Lilly, Indianapolis, Ind) was administered as a split dose of 0.5 mg just prior to single-shot fast SE and gadolinium-enhanced imaging. Transverse and coronal single-shot fast SE MR images (⬁/94 [repetition time

msec/echo time msec], 256 ⫻ 256 matrix, one-half signal acquired, 7-mm section thickness with 3-mm gap, echo train length of 94, receiver bandwidth of ⫾62 kHz) of the entire abdomen and pelvis were acquired during breath hold. Noninterleaved sets of 12 sections were obtained during each 19-second breath hold. Typically, four breath holds were required to image the abdomen and pelvis in the transverse plane. For the spoiled GRE breath-hold examinations with fat saturation, transverse images through the entire abdomen and pelvis were acquired immediately after intravenous administration of 0.2 mmol of gadodiamide (Omniscan; Amersham Health, Princeton, NJ) per kilogram of body weight as a bolus at a rate of 2 mL/sec. To cover the abdomen and pelvis in the transverse plane, four 24-second breath holds were required, each producing 12 noninterleaved sections. Coronal and delayed transverse spoiled GRE images (141–165/2.1, 512–256 ⫻ 192 matrix, three-quarters field of view for the transverse acquisition, one signal acquired, 10-mm section thickness with no intersection gap, ⫾16 –20-kHz receiver bandwidth, 70° flip angle) were obtained 5 minutes after contrast material injection. A minimum out-of-phase echo time of 2.1 msec was chosen to accentuate the effects of chemical fat suppression on the GRE images.

Blinded Separate Review of Single-shot Fast SE and Gadolinium-enhanced Spoiled GRE Images Two radiologists (R.N.L., C.P.S.) blinded to the results of endoscopy, barium studies, cross-sectional imaging examinations, and patient identity randomly and independently evaluated single-shot fast SE and gadolinium-enhanced fat-suppressed spoiled GRE MR images. Groups of the patient studies were mixed, and both types of images were reviewed by each observer. MR images were evaluated for mural thickening and abnormal enhancement of the stomach, duodenum, jejunum, ileum, ascending colon, transverse colon, descending colon, or rectosigmoid colon. In each patient, abnormal bowel segments were indicated for each image type.

Definitions Bowel wall enhancement was considered to be abnormal if the degree of enhancement was greater than that of liver parenchyma at visual inspection. Mild mural enhancement was noted if the en-

MR Imaging of Crohn Disease: Endoscopic Correlation

䡠

653

Figure 1. Images obtained through the lower pelvis in a 48-year-old woman with an acute exacerbation of Crohn disease. (a) Transverse breath-hold single-shot fast SE MR image (⬁/94, 90° flip angle) depicts a normal-appearing sigmoid colon without definite mural thickening. Note the mural thickening (short arrow) at the junction of the descending and sigmoid colon and the moderate mural thickening of the distal ileum (long arrows). (b) Transverse fat-suppressed gadolinium-enhanced spoiled GRE MR image (165/2, 70° flip angle) shows abnormal mural thickening and marked enhancement of the sigmoid colon (long arrows) and rectum. Similar changes are present in the terminal ileum (short arrow). Findings correlate with moderate to marked active Crohn ileitis and colitis.

hancement was greater than that of liver parenchyma but lower than that of intravascular enhancement. Marked mural enhancement was noted if the degree of enhancement was equal to intravascular signal enhancement. In patients with fatty infiltration of the liver, the signal intensity of the spleen was used as an internal standard for normal enhancement on delayed images. Bowel wall thickness of 3 mm or less was considered normal. Mild thickening was noted for bowel thickness of 4– 6 mm; moderate, for wall thickness of 7–12 mm; and marked, for wall thickness greater than 12 mm. Thickness was determined by means of visual inspection and with use of calipers. Complications of Crohn disease including abscess, phlegmon, fistulas, and associated biliary disease were also recorded for each pulse sequence.

Conspicuity of the Diseased Bowel Single-shot fast SE and gadolinium-enhanced spoiled GRE images were reviewed side by side. The conspicuity of the bowel wall findings was scored by using the following three–point scale: 1, not visualized; 2, findings were poorly seen and difficult to distinguish from normal bowel; and 3, abnormal bowel was well visualized with no difficulty distinguishing diseased from normal bowel.

Extent of Disease For each patient, the extent of disease was determined with establishment of 654

䡠

Radiology

䡠

March 2002

the number of individual bowel segments that were involved with Crohn disease. The extent of Crohn disease on single-shot fast SE and gadolinium-enhanced spoiled GRE MR images, as determined from the blinded review of MR images, was compared with that determined at endoscopy, barium studies, and surgery. A segment-by-segment correlation was made between the imaging studies and the proof of disease at endoscopy, surgery, and barium examinations. Results for each segment were categorized as true-positive, false-negative, truenegative, or false-positive. Bowel segments with disease that could not be proven were excluded from consideration. For single-shot fast SE and gadolinium-enhanced spoiled GRE images, the observers determined whether the two types of MR images correctly depicted or caused overestimation or underestimation of the extent of Crohn disease. In some patients, there were both false-negative and false-positive interpretations for the involvement of the bowel segment. In these cases, the predominant error was chosen to characterize the case as underestimating or overestimating the extent of disease.

Severity of Disease Clinical findings about the severity of Crohn disease were compared with MR findings. The severity was determined from discussions with each patient’s gastroenterologist and from review of endo-

scopic reports and histopathologic findings. Crohn disease severity was scored as (a) no disease, if there was no endoscopic or histopathologic evidence of disease; (b) mild disease, if only mild mucosal changes with erythema were present at endoscopy and histopathologic findings showed active inflammation; (c) moderate to marked disease, if there were more severe endoscopic mucosal changes with ulcerations, friable mucosa, and/or fistulas, and biopsy showed active inflammation; and (d) chronic inactive disease, if endoscopic findings showed no active mucosal disease but biopsy findings showed evidence of chronic inflammation. Evaluation of the severity of disease was performed by means of consensus of the two observers. For each patient, information obtained from MR images of the most severely diseased bowel segments was used to score overall disease severity. Disease severity was categorized as (a) no disease, (b) mild active disease, (c) moderate to severe active disease, and (d) chronic inactive disease. For singleshot fast SE images, the following degree of mural thickening and wall signal intensity were used to estimate disease severity: (a) normal bowel wall thickness, no disease; (b) mild mural thickening, mild disease; (c) moderate to marked mural thickening, moderate to marked disease; and (d) mural thickening with a markedly hypointense wall (intensity equal to or less than that of the psoas muscle), chronic disease. For gadolinium-enhanced spoiled GRE MR images, a Low et al

Figure 2. Images obtained in a 54-year-old man with Crohn disease who presented with recurrent lower abdominal pain after resection of the terminal ileum. (a) Coronal single-shot fast SE MR image (⬁/90°, 90° flip angle) shows the colon distended with rectal water. The ileocolonic anastomosis (arrow) is unremarkable. (b) Coronal gadoliniumenhanced spoiled GRE MR image (165/2.1, 70° flip angle) shows focal mural thickening and enhancement (arrow) at the ileocolonic anastomosis. (c) Colonoscopic image confirms focal ulceration (arrow) at the ileocolonic anastomosis.

combination of the degree of mural thickening and mural enhancement on the first set of transverse images was used to estimate Crohn disease severity: (a) no mural thickening and no enhancement, no disease; (b) mural thickening and mild enhancement, mild disease; (c) moderate or marked mural thickening and marked enhancement, moderate to marked disease; and (d) any degree of mural thickening and no enhancement, inactive chronic disease.

Statistical Analysis For the depiction of mural changes in Crohn disease on single-shot fast SE and gadolinium-enhanced spoiled GRE imVolume 222

䡠

Number 3

ages, sensitivity, specificity, and accuracy were calculated on a per-patient and perbowel segment basis. The images were compared by using the McNemar test of correlated proportions with the Yates continuity correction. Two-tailed P values and ␹2 values were reported. The null hypothesis was rejected for P ⬍ .05. The ␬ coefficient of interobserver agreement was calculated for the presence or absence of abnormal bowel segments on single-shot fast SE and gadolinium-enhanced spoiled GRE images.

RESULTS Twenty-five of 28 patients had abnormal bowel segments proven at endoscopy,

barium studies, or surgery (Fig 1); three patients had no evidence of active or chronic Crohn disease. Gadolinium-enhanced spoiled GRE MR images that depicted abnormal bowel segments in 25 (observer 1) and 24 (observer 2) of 28 patients had a per-patient sensitivity of 100% and 96%, respectively (Fig 2). In comparison, single-shot fast SE MR images had a per-patient sensitivity of 60% for both observers (P ⬍ .05) (McNemar ␹2 ⫽ 5.81 and 4.90) for depicting diseased bowel segments in 15 of 25 patients. The corresponding per-patient specificity was 67% for both observers for gadoliniumenhanced spoiled GRE images, compared with 33% and 67% for single-shot fast SE images. The per-patient accuracy was 96% and 93% for gadolinium-enhanced spoiled GRE images compared with 57% and 61% (P ⬍ .05) (McNemar ␹2 ⫽ 5.81) for single-shot fast SE images.

Depiction of Abnormal Bowel Segments In the segment-by-segment analysis, 205 segments were evaluated. Nineteen segments were excluded in patients who had prior bowel resections or in areas in

MR Imaging of Crohn Disease: Endoscopic Correlation

䡠

655

Figure 3. Images obtained through the pelvis in a 37-year-old man with recently diagnosed Crohn disease. (a) Transverse breath-hold single-shot fast SE MR image (⬁/94, 30° flip angle) is unremarkable. Rectal water distends the rectosigmoid colon without evidence of mural thickening. (b) Transverse fat-suppressed gadolinium-enhanced spoiled GRE MR image (165/2, 70° flip angle) shows mild mural thickening and marked enhancement of the rectosigmoid colon (arrows). Similar changes are noted throughout the colon and at the terminal ileum. (c) Colonoscopic image shows scattered linear ulcerations (arrows) involving the entire colon and the terminal ileum.

GRE images, which depicted disease extent in 18 and 17 patients for the two observers compared with single-shot fast SE images, which depicted disease in nine patients.

Severity of Disease

which there was no proof of disease. A total of 61 segments were involved with Crohn disease. For the two observers, gadolinium-enhanced spoiled GRE MR images depicted more segments of the diseased bowel in 16 and 15 of the 28 patients, while in the remaining patients, the 12 and 13 patients, gadolinium-enhanced spoiled GRE images and singleshot fast SE images depicted an equal number of segments involved with Crohn disease. In no patient did the single-shot fast SE images depict more abnormal bowel segments than did gadolinium-enhanced spoiled GRE images. Overall, for the two observers, gadolinium-enhanced spoiled GRE MR images depicted 54 and 52 of the 61 segments of the diseased bowel compared with singleshot fast SE MR images, which depicted 31 and 32 of the segments (Fig 3). Gadolinium-enhanced spoiled GRE MR images had a per-segment sensitivity of 89% and 656

䡠

Radiology

䡠

March 2002

85%, specificity of 96% and 94%, and accuracy of 94% and 91%, compared with single-shot fast SE MR images sensitivity of 51% and 52% (P ⬍ .001) (McNemar ␹2 ⫽ 18.37 and 15.43), specificity of 98% and 96%, and accuracy of 84% and 83% (P ⬍ .01) (McNemar ␹2 ⫽ 12.89 and 9.48) (Table 1).

Extent of Disease Extent of disease was defined as the number of bowel segments involved with Crohn disease in each patient. Table 2 compares gadolinium-enhanced spoiled GRE images and single-shot fast SE images with the results of endoscopy, barium studies, and surgery in the prediction of the extent of gastrointestinal tract disease in the 28 patients. The extent of the involvement of the gastrointestinal tract with Crohn disease was better depicted on gadolinium-enhanced spoiled

Disease severity, as determined from clinical, endoscopic, and histopathologic findings, was no disease in three patients, inactive chronic disease in two patients, mild disease in three patients, and moderate to marked disease in the remaining 20 patients (Fig 4). Gadolinium-enhanced spoiled GRE MR images were superior for determining the severity of Crohn disease (Fig 4). The images helped to correctly estimate the severity of disease in 26 (93%) of 28 patients, while causing underestimation in one (4%) patient and overestimation in one (4%) patient. Single-shot fast SE images helped to correctly estimate the severity of Crohn disease in 12 (43%) of 28 patients, while causing underestimation in 13 (46%) patients and overestimation in three (11%) patients. Disease activity correlated with the degree of mural enhancement on gadolinium-enhanced spoiled GRE images (Fig 5). Mural thickening without gadolinium enhancement was seen in two patients with chronic or inactive Crohn disease (Fig 6). On single-shot fast SE Low et al

TABLE 1 Depiction of the Diseased Bowel Segment: Comparison of Single-Shot Fast SE and Gadolinium-enhanced Spoiled GRE MR Images Images Single-shot fast SE Observer 1 Observer 2 Gadolinium-enhanced spoiled GRE Observer 1 Observer 2

True-Positive

False-Negative

True-Negative

False-Positive

Sensitivity (%)

Specificity (%)

Accuracy (%)

31 32

30 29

141 138

3 6

51 52

98 96

84 83

54 52

7 9

138 135

6 9

89* 85*

96 94

94† 91†

Note.—Unless otherwise noted, data are the number of bowel segments involved with Crohn disease in 28 patients. * P ⬍ .001. † P ⬍ .01.

TABLE 2 Determination of the Extent of Crohn Disease: Comparison of Single-Shot Fast SE and Gadolinium-enhanced Spoiled GRE MR Images Images Single-shot fast SE Observer 1 Observer 2 Gadolinium-enhanced spoiled GRE Observer 1 Observer 2

Correct Estimate

Overestimate

Underestimate

9 (32) 9 (32)

3 (11) 3 (11)

16 (57) 16 (57)

19 (68) 17 (61)

6 (21) 6 (21)

3 (11) 5 (18)

Note.—Data are the number of patients. Twenty-eight patients had bowel segments involved with Crohn disease. Data in parentheses are percentages.

images, it was not possible to distinguish chronic from active disease, as both showed mural thickening. On these images, the signal intensity of the wall of the bowel was comparable with that of the normal bowel, actively diseased bowel, and bowel segments with chronic inactive Crohn disease.

Conspicuity of Mural Changes Gadolinium-enhanced spoiled GRE MR images showed consistently greater conspicuity of abnormal bowel segments compared with single-shot fast SE images (Fig 5). The mean score for the conspicuity of diseased bowel segments was 2.88 ⫾ 0.33 (mean ⫾ SD) for gadoliniumenhanced spoiled GRE images compared with 1.88 ⫾ 0.83 for single-shot fast SE images. In 17 of 25 patients with diseased bowel segments, gadolinium-enhanced images were given a higher score for disease conspicuity compared with singleshot fast SE MR images. Complications of Crohn disease were depicted equally on both types of images, which correctly depicted fistulas in two patients and an abdominal abscess in one patient. The ␬ analysis showed good interobVolume 222

䡠

Number 3

server agreement for the depiction of abnormal bowel segments in patients with Crohn disease. Overall, the two observers agreed in 93% of determinations, ␬ ⫽ 0.79.

DISCUSSION Recent advances in MR imaging hardware and software have dramatically improved image quality while simultaneously decreasing imaging times. Highperformance gradient systems and faster pulse sequences with more efficient schemes for acquiring k-space data have made T1- and T2-weighted breath-hold MR imaging feasible. Many of the challenges of abdominal MR imaging that relate to physiologic motion have been overcome with these breath-hold image acquisitions, which have become a routine part of abdominal MR imaging examinations at our institution. MR imaging of the gastrointestinal tract has been particularly challenging due to bowel peristalsis, respiratory motion, and lack of readily available intraluminal contrast agents. Prior reports (6,8 – 12,16,19,25) have described MR imaging of Crohn disease with use of a combina-

tion of unenhanced and gadolinium-enhanced imaging. While evaluation of the anorectal complications of Crohn disease is a recognized strength of MR imaging (22–24), evaluation of the remainder of the gastrointestinal tract has been limited by motion artifact. With faster breath-hold pulse sequences, MR evaluation of the Crohn disease involvement of the colon and small bowel has also become feasible. In this study, we evaluated one approach to MR imaging of Crohn disease with T2-weighted breath-hold single-shot fast SE and T1-weighted gadolinium-enhanced spoiled GRE MR imaging combined with readily available and inexpensive intraluminal contrast agents. The efficiency of breath-hold imaging allows completion of the abdominal and pelvic MR examinations in 20 minutes. In our study, gadolinium-enhanced spoiled GRE MR images were most useful for depicting the wall of the inflamed bowel segments and for determining the extent of disease. The superior performance of these images, compared with single-shot fast SE images, for depicting mural changes of Crohn disease was due to the marked enhancement of the diseased bowel segment with use of intravenous gadolinium-based contrast agent. Dilute barium sulfate and water distend the bowel lumen and serve as negative intraluminal contrast agents on T1weighted spoiled GRE images. Enhancement of the inflamed segments of the bowel in patients with Crohn disease is easily depicted when the surrounding extraintestinal fat is suppressed and the bowel lumen is filled with a negative contrast agent. Disease severity was also best determined on gadolinium-enhanced spoiled GRE images, which helped to correctly predict the severity of Crohn disease in 93% of

MR Imaging of Crohn Disease: Endoscopic Correlation

䡠

657

Figure 4. Images obtained in a 30-year-old woman who presented with severe abdominal pain and diarrhea. (a) Transverse single-shot fast SE MR image (⬁/90, 90° flip angle) shows the colon (arrows). Administration of rectal water was limited by the patient’s acute perirectal symptoms. Marked mural thickening is present but is difficult to distinguish from intraluminal stool. (b) Transverse gadolinium-enhanced fat-suppressed spoiled GRE MR image (165/2.1, 70° flip angle) shows pancolitis (arrows) with marked mural thickening and enhancement. Severe luminal narrowing is present. (c) Colonoscopic image confirms marked changes in the Crohn disease with edematous, inflamed, and friable mucosa. Heaped up folds of abnormal mucosa produce a cobblestone appearance, which results in luminal narrowing.

patients, compared with single-shot fast SE images, which depicted disease in 43% of patients. This assessment was limited by the fact that most of our study patients had active disease. In our study, however, disease severity correlated best with the degree of mural enhancement with use of intravenous gadolinium-based contrast agent at the first gadolinium-enhanced acquisition. Mural thickening without enhancement correlated with chronic or inactive Crohn disease. Mural thickening with mild enhancement was seen in patients with mild active Crohn disease, and mural thickening with marked enhancement correlated with moderate to severe active Crohn disease. On the other hand, mural enhancement without any mural thickening should be interpreted with caution, as such cases resulted in several falsepositive interpretations. It is now our practice to describe the 658

䡠

Radiology

䡠

March 2002

diseased bowel when mural enhancement is accompanied by at least mild mural thickening. False-positive interpretations for the presence of Crohn involvement of an individual bowel segment occurred with both single-shot fast SE images and gadolinium-enhanced spoiled GRE images. All but two patients had other segments with proven mural changes of Crohn disease that were correctly depicted on MR images. Half-Fourier MR imaging implemented as single-shot fast SE MR imaging has been proposed in the evaluation of bowel disease (17–21). This breathing-independent T2-weighted acquisition uses halfFourier transformation and a very long echo train (⬎90) to efficiently fill k space, which facilitates T2-weighted breathhold imaging. Since each section is acquired independently in less than 1 second, physiologic motion due to bowel

peristalsis is eliminated. Compared with spoiled GRE MR images, single-shot fast SE images are also less sensitive to susceptibility artifacts that arise from bowel gas. The conspicuity of a gastrointestinal mural abnormality is determined by the relative contrast enhancement between the mural thickening and the adjacent fat outside the bowel and the intraluminal contrast material within the bowel lumen. Both water and dilute 2% barium sulfate serve as positive intraluminal contrast agents on single-shot fast SE images. These images are useful for depicting bowel lumen that is distended with highsignal-intensity water. In our study, we found that on single-shot fast SE images, the intermediate signal intensity of the bowel wall was depicted clearly adjacent to the high signal intensity of the surrounding fat and that of the intraluminal water. Some authors have advocated the use of fat suppression at T2-weighted MR imaging in patients with Crohn disease, which allows evaluation of the increased signal intensity of the wall of the diseased bowel segments (10). The addition of a fat-suppressed single-shot fast SE acquisition could be useful in the depiction of inflammatory changes in the surrounding soft tissues and would result in Low et al

Figure 5. Images obtained in a 58-year-old woman with Crohn disease who presented with right-sided abdominal pain after left hemicolectomy and transverse colostomy. (a) Transverse breath-hold single-shot fast SE MR image (⬁/94, 90° flip angle) shows dilute barium distending the right colon without definite mural thickening. Mild thickening of the transverse colon (arrow) is present at the colostomy site. (b) Transverse fat-suppressed gadolinium-enhanced spoiled GRE MR image (165/2, 70° flip angle) shows mild to moderate right colonic mural thickening and marked enhancement (arrows) accentuated by the negative intraluminal contrast material.

Figure 6. Images obtained in a 50-year-old woman with Crohn disease and prior surgical resection of the terminal ileum and cecum. The patient presented with recurrent lower abdominal pain and diarrhea. (a) Transverse single-shot fast SE MR image (⬁/94, 90° flip angle) shows moderate mural thickening (arrow) and stenosis of the distal ileum. (b) Transverse gadolinium-enhanced fat-suppressed spoiled GRE MR image (165/2.1, 70° flip angle) depicts moderate mural thickening of the distal ileum (arrow) with no substantial mural enhancement. Findings correlate with chronic stricture. Colonoscopy helped to confirm a chronic stricture without active inflammation.

only a slight increase in the examination time. On T2-weighted fat-suppressed images, depiction of the bowel wall is more effective with a negative intraluminal contrast agent, such as iron oxide, that contains oral agents. With a positive intraluminal contrast agent, non–fat-suppressed single-shot fast SE images allow better visualization of the bowel wall and evaluation of its thickness. The high intraluminal signal intensity from water would make it more difficult to define Volume 222

䡠

Number 3

the adjacent bowel wall on T2-weighted fat-suppressed images. Our results concur with those of Marcos and Semelka (25), who, in a preliminary comparison of half-Fourier RARE and gadolinium-enhanced spoiled GRE images of Crohn disease, found that the severity of disease extent could not be evaluated with half-Fourier RARE images, as the signal intensity of the diseased bowel was comparable to that of the normal bowel in 10 of 11 patients. On gad-

olinium-enhanced spoiled GRE images, mural enhancement with gadoliniumbased contrast agent correlated with disease severity in 10 of 11 patients. The use of MR imaging in the evaluation of Crohn disease activity and its response to treatment have also been reported. Some authors note that the results of gadolinium-enhanced MR imaging may be a more accurate predictor of Crohn disease activity than the Crohn disease activity index (10,16). Maccioni

MR Imaging of Crohn Disease: Endoscopic Correlation

䡠

659

et al (10) evaluated the biologic activity, as measured with three acute-phase reactants, in patients with Crohn disease. They found that disease activity showed a substantial correlation with gadolinium enhancement of the diseased bowel wall, wall hyperintensity on T2-weighted fatsuppressed images, and hyperintensity of fibrofatty proliferation on T2-weighted fat-suppressed images. Madsen et al (26) evaluated the treatment response in patients with active Crohn disease and found that during treatment, substantial decrease of T2-weighted signal intensity, mural gadolinium enhancement, and bowel wall thickness was noted. The important distinction between an actively inflamed stenosis and a chronically scarred fibrotic stenosis correlates with the degree of mural gadolinium enhancement (16) (Fig 4). Limitations of our study should be acknowledged. Use of a phased-array surface coil would have increased image signal-to-noise ratio, which might have improved the performance of single-shot fast SE images. However, our examination of the abdomen and pelvis included 48 cm in a craniocaudal direction. At the time of this study, surface coils available for our imager had a much more limited coverage. The inhomogeneous image quality of phased-array coil images is also problematic. Common features are areas of signal flash at the anterior and signal drop-off at the center of the images obtained with phased-array coils. This image inhomogeneity makes interpretation of subtle bowel disease and enhancement extremely difficult. For these reasons, we choose the body coil for combined abdominal and pelvic MR examinations. All 28 patients had proven Crohn disease, and most patients had active disease. Examination of more patients with chronic inactive disease would be useful. However, our conclusions regarding the correlation of disease severity and the degree of gadolinium enhancement are supported by results of prior studies (6, 10,26). The application of these MR imaging techniques to other types of bowel disease is not addressed in our study. The findings of bowel wall enhancement and thickening are clearly nonspecific. Other infectious, inflammatory, or neoplastic diseases affecting the bowel wall will demonstrate similar MR findings. In patients with suspected Crohn disease, endoscopy and biopsy are still required to

660

䡠

Radiology

䡠

March 2002

establish the diagnosis. Further studies in which our MR imaging technique is compared with helical CT will be required. In conclusion, fat-suppressed gadolinium-enhanced spoiled GRE MR imaging is more effective than single-shot fast SE MR imaging in the evaluation of mural gastrointestinal changes in Crohn disease. The use of oral dilute 2% barium sulfate and rectal water distends the bowel lumen and serves as positive contrast agent on single-shot fast SE images and as a negative contrast agent on gadolinium-enhanced spoiled GRE images. Enhancement of the diseased segments of the bowel on fat-suppressed gadolinium-enhanced spoiled GRE MR images facilitates depiction of the extent and severity of Crohn disease.

12.

13.

14.

15.

16.

17. References 1. Farmer RG, Hawk WA, Turnbull RB. Clinical patterns in Crohn’s disease: a statistical study of 615 cases. Gastroenterology 1975; 68:627– 635. 2. Wills JS, Lobis IF, Denstman FJ. Crohn disease: state of the art. Radiology 1997; 202:597– 610. 3. Goldberg HI, Gore RM, Margulis AR, Moss AA, Baker EL. Computed tomography in the evaluation of Crohn disease. AJR Am J Roentgenol 1982; 140:277–282. 4. Gore RM, Balthazar EJ, Ghahremani GG, Miller FH. CT features of ulcerative colitis and Crohn’s disease. AJR Am J Roentgenol 1996; 167:3–15. 5. Chou CK, Chen LT, Sheu RS, Wang ML, Jaw TS, Liu GC. MRI manifestations of gastrointestinal wall thickening. Abdom Imaging 1994; 19:389 –394. 6. Kettritz U, Isaacs K, Warshauer DM, Semelka RC. Crohn’s disease: pilot study comparing MRI of the abdomen with clinical evaluation. J Clin Gastroenterol 1995; 21:249 –253. 7. Mirowitz SA. Contrast enhancement of the gastrointestinal tract on MR images using intravenous gadolinium-DTPA. Abdom Imaging 1993; 18:215–219. 8. Low RN, Francis IR, Politoske D, Bennett M. Crohn’s disease evaluation: comparison of contrast-enhanced MR imaging and single phase helical CT scanning. J Magn Reson Imaging 1999; 11:127–135. 9. Low RN, Francis IR. MR imaging of the gastrointestinal tract with IV gadolinium and diluted barium oral contrast media compared with unenhanced MR imaging and CT. AJR Am J Roentgenol 1997; 169: 1051–1059. 10. Maccioni F, Viscido A, Broglia L, et al. Evaluation of Crohn disease activity with magnetic resonance imaging. Abdom Imaging 2000; 25:219 –228. 11. Semelka RC, Shoenut JP, Silverman R, Kroeker MA, Yaffe CS, Micflikier AB. Bowel disease: prospective comparison of CT and 1.5-T pre- and postcontrast MR imaging with T1-weighted fat-suppressed

18.

19.

20.

21.

22.

23.

24.

25.

26.

and breath-hold FLASH sequences. J Magn Reson Imaging 1991; 1:625– 632. Anderson CM, Brown JJ, Balfe DM, et al. MR imaging of Crohn disease: use of perflubron as a gastrointestinal contrast agent. J Magn Reson Imaging 1994; 4:491– 496. Brown JJ, Duncan JR, Heiken JP, et al. Perfluoroctylbromide as a gastrointestinal contrast agent for MR imaging: use with and without glucagon. Radiology 1991; 181:455– 460. Rubin DL, Muller HH, Nino-Murcia M, et al. Intraluminal contrast enhancement and MR visualization of the bowel wall: efficacy of PFOB. J Magn Reson Imaging 1991; 1:371–380. Rijcken THP, Davis MA, Ros PR. Intraluminal contrast agents for MR imaging of the abdomen and pelvis. J Magn Reson Imaging 1994; 4:291–300. Schunk K, Kern A, Oberholzer K, et al. Hydro-MRI in Crohn’s disease: appraisal of disease activity. Invest Radiol 2000; 35: 431– 437. Marcos HB, Semelka RC. Stomach diseases: MR evaluation using combined T2weighted single-shot echo train spinecho and gadolinium-enhanced spoiled gradient-echo sequences. J Magn Reson Imaging 1999; 10:950 –960. Regan F, Beall DP, Bohlman ME, Khazan R, Sufi A, Schaefer DC. Fast MR imaging and the detection of small-bowel obstruction. AJR Am J Roentgenol 1998; 170: 1465–1469. Ernst O, Asselah T, Casblan X, Sergent G. Breath-hold fast spin-echo MR imaging of Crohn’s disease. AJR Am J Roentgenol 1998; 170:127–128. Chung JJ, Semelka RC, Martin DR, Marcos HB. Colon diseases: MR evaluation using combined T2-weighted single-shot echo train spin-echo and gadolinium-enhanced spoiled gradient-echo sequences. J Magn Reson Imaging 2000; 12:297–305. Marcos HB, Semelka RC, Noone TC, Woosley JT, Lee JK. MRI of normal and abnormal duodenum using half-Fourier single-shot RARE and gadolinium-enhanced spoiled gradient echo sequences. Magn Reson Imaging 1999; 17:869 – 880. O’Donovan AN, Somers S, Farrow R, Mernagh JR, Sridhar S. MR imaging of anorectal Crohn disease: a pictorial essay. RadioGraphics 1997; 17:101–107. Laniado M, Makowiec F, Dammann F, Jehle EC, Claussen CD, Starlinger M. Perianal complications of Crohn disease: MR imaging findings. Eur Radiol 1997; 7:1035–1042. Haggett PJ, Moore NR, Shearman JD, Travis SP, Jewell DP, Mortensen NJ. Pelvic and perineal complications of Crohn’s disease: assessment using magnetic resonance imaging. Gut 1995; 36:407– 410. Marcos HB, Semelka RC. Evaluation of Crohn’s disease using half-Fourier RARE and gadolinium-enhanced SGE sequences: initial results. Magn Reson Imaging 2000; 18:263–268. Madsen SM, Thomsen HS, Schlichting P, Dorph S, Munkholm P. Evaluation of treatment response in active Crohn’s disease by low-field magnetic resonance imaging. Abdom Imaging 1999; 24:232–239.

Low et al