MRI of the Gastrointestinal System

▪ Introduction

Although computed tomography (CT) has been and still is the mainstay for imaging the gastrointestinal (GI) tract, small bowel and colorectal magnetic resonance (MR) applications have been developed and increasingly adopted in recent years. MR enterography evaluates the small bowel for inflammatory processes, such as Crohn’s disease, neoplasms, and etiologies of obstruction and bleeding. Colorectal applications include rectal cancer staging, anal fistula evaluation, and appendicitis (usually in the setting of pregnancy or in the pediatric population).

▪ Small Bowel

CT and MRI complement one another in imaging the small bowel. Whereas CT features lower cost, convenience, and availability, MRI obviates radiation exposure, iodinated contrast, and provides superior tissue contrast and multiparametric evaluation (signal characteristics, enhancement, diffusion restriction, and peristalsis). Its chief role is to avoid or minimize exposure to ionizing radiation in the setting of conditions presenting early in life and requiring repetitive imaging (ie, Crohn’s disease and polyposis syndromes). Other indications for MR imaging include: other inflammatory conditions, small bowel masses, relative CT contraindications (ie, pregnancy), to identify the etiology of small bowel obstruction and GI bleeding in certain cases, and after an incomplete capsule endoscopy ( Box 8.1 ).

BOX 8.1

Indications for Magnetic Resonance (MR) Enterography Inflammatory Etiologies

Crohn’s disease

Celiac disease

Bowel ischemia and vasculitis

Radiation- and chemotherapy-induced enteritis

Small Bowel Neoplasms

Diagnosis and follow-up of small bowel tumors

Peritoneal ingrowth of metastatic disease

Polyposis Syndromes

Peutz-Jeghers syndrome

Juvenile polyposis

Cowden syndrome

Gardner syndrome

CTE Contraindications

Pregnancy

Iodinated contrast allergy

Inability to tolerate oral contrast

Radiation dose considerations

Miscellaneous

Small bowel obstruction

Gastrointestinal bleeding after capsule endoscopy

▪ Technical Considerations

Optimal imaging requires small bowel distention, contrast enhancement, rapid imaging, and prone positioning to preempt artifacts from bulk motion, peristaltic motion, and susceptibility. Even though MR enteroclysis outperforms MR enterography in distending small bowel loops and demonstrating luminal abnormalities, technical considerations and patient comfort generally mitigate in favor of MR enterography. In MR enterography, adequate bowel distention is achieved with the administration of a large volume of an oral contrast agent. Oral contrast agents fall into three broad categories based on their imaging appearance: 1) negative (T1- and T2-hypointense), 2) positive (T1- and T2-hyperintense), and 3) biphasic (T1-hypointense and T2-hyperintense) ( Fig. 8.1 ). Biphasic agents present the best tissue contrast scenarios—T1-hypointensity against hyperenhancement and T2-hyperintensity against relatively hypointense bowel wall. A variety of dosing regimens are prescribed, usually involving up to 2 L of contrast administered during the hour preceding the examination. The protocol at our institution calls for one bottle of barium sulfate every 20 minutes for a total of three bottles, or 1350 mL of barium sulfate.

In additional to fasting (at least 2 hours at our institution) antiperistaltic agents provide the opportunity to minimize image degradation as a result of bowel peristaltic activity. In the United States, antiperistaltic options include glucagon and hyoscyamine (Levsin) (butylscopolamine is not FDA-approved). Intramuscular and intravenous glucagon and sublingual and intravenous hyoscyamine formulations provide a number of options, but all complicate workflow. Although image quality suffers more from motion artifact without antiperistaltic use, the advantages potentially justify abandoning their use. The small magnitude of the hyoscyamine effect, the potential benefit of highlighting inflamed bowel segments (the “frozen bowel sign”), the preemption of medication side effects, and the decreased cost and streamlined workflow all favor at least considering abandoning antiperistaltic administration. However, in the setting of tumor identification and/or when assessment of the “frozen bowel” sign is not relevant, minimizing peristalsis and its attendant artifacts and image degradation offers more relative benefit.

As with other body MRI applications, MR enterography necessitates the use of a dedicated torso coil. If possible, prone positioning offers a number of advantages over supine positioning: 1) faster imaging in the coronal plane because of compressive effects, 2) better bowel distention, and 3) elimination of abdominal wall motion artifact. Intravenous contrast enhancement is critical to assess the acuity/chronicity of inflammation and its complications, to help identify and characterize masses and highlight associated findings, such as vascular engorgement and surrounding inflammation and neoplastic spread. Although the higher relaxivity of gadobenic acid (MultiHance ^® ) recommends its use in MR enterography, recent reports suggesting gadolinium accumulation in the brain (irrespective of renal function) argue against its use. Although the clinical significance of this phenomenon is potentially nil, contrast enhancement with a macrocyclic agent circumvents this potential problem. As with all other body MRI applications, dynamic postcontrast imaging (as described in Chapter 1 ) is recommended as a means of discriminating between acute and chronic inflammation and characterizing tumors.

Pulse sequence parameters differ from other applications in a number of ways ( Table 8.1 ). The distribution of small bowel loops favors coronal plane prescription, especially for time-sensitive pulse sequences, such as the dynamic sequence, and a combination of axial and coronal sequences are acquired. The MR enterography pulse sequences conform to the protocol scheme presented in Chapter 1 pivoting on the T1- versus T2-weighted framework. Steady-state images provide a nice “T2-weighted” (really T2/T1-weighted) overview, courtesy of rapid imaging, insensitivity to motion artifacts, and fluid sensitivity. Single-shot heavily T2-weighted images feature similar attributes, although suffer from motion artifact related to intraluminal motion artifact ( Fig. 8.2 ). Fat suppression improves tissue contrast and dynamic range, but compromises signal-to-noise ratio (SNR) in relatively SNR-poor single-shot images relying on a single excitation pulse per image. (Fast spin-echo [FSE] images applied for abdominal visceral imaging suffer prohibitively from peristaltic motion artifact because of the higher acquisition time.) In addition to the inherent T2 contrast, diffusion-weighted imaging (DWI) helps to isolate inflammation and neoplastic hypercellularity ( Fig. 8.2 ). Finally, motility imaging implementing rapid T2-weighted—usually steady-state—pulse sequences with multiple frames per slice location demonstrates peristaltic activity (and its absence in the setting of inflammation).

TABLE 8.1

Sample Magnetic Resonance (MR) Enterography Protocol

Pulse Sequence	Relevant Parameters	Utility	Limitations
Coronal (or 3-plane) steady-state	6 × 0 mm slice thickness	Fluid-solid tissue contrast; motion insensitive	Prone to susceptibility and banding (moiré) artifacts
Coronal SSFSE	TE ≅ 200 msec	Fluid sensitivity; motion and susceptibility artifact insensitive	Poor SNR further compromised with fat suppression; intraluminal fluid motion artifact
Axial SSFSE	TE ≅ 200 msec	Same as above	Same as above
Coronal in- and out-of-phase	Ideally derived from Dixon dynamic sequence	Mesenteric changes; incidental findings (ie, hepatic steatosis, adrenal adenoma)	Minimal bowel tissue contrast
Dynamic	3-D fat-suppressed with bolus-timing	Detect and characterize inflammation and tumors	Bowel wall blurring
Axial fat-suppressed FSE	TE ≅ 80 msec	Bound-water/bowel wall tissue contrast	Motion artifact
Coronal fat-suppressed FSE	TE ≅ 80 msec	Same as above	Same as above
Coronal delayed	Same as dynamic	Adds contrast kinetic information (accentuates extracellular tissues, such as inflammation)	Same as dynamic
Axial delayed	3-D with fat suppression	Same as above	Same as above
DWI	b = 800	Extreme tissue contrast and sensitivity to inflammation and neoplasms	Prone to artifacts
Coronal cine steady-state	Approximately 10 slice locations with ≅ 25 phases per location	Characterize peristalsis	Breathing motion artifact

The primary utility of T1-weighted images is to illustrate contrast enhancement. Dynamic fat-suppressed T1-weighted 3-dimensional (3-D) images are obtained in the coronal plane to meet time constraints as previously discussed, whereas delayed fat-suppressed T1-weighted 3-D image acquisition is performed axially and/or coronally ( Fig. 8.3 ). Although 3-D images suffer from bowel wall blurring from peristaltic activity, the higher slice resolution, lack of respiratory misregistration, and lack of time-of-flight pseudoenhancement (from motion) favor 3-D over 2-D acquisition. Adding in- and out-of-phase images (and fat images when dynamic sequences are performed with Dixon technique as discussed in Chapter 1 ) provides an anatomic overview and another means of detecting mesenteric inflammation and tumor spread.

▪ Normal Appearance

Normal small bowel diameter measures less than 3 cm in diameter and varies in MR enterography studies, depending on the degree of oral contrast distention. The bowel wall thickness also varies in proportion to the degree of oral contrast distention and generally measures less than 3 mm in thickness. Bowel wall and fold signal intensity is uniformly hypointense and enhancement is minimal. Underdistention can simulate pathology with relative wall and fold thickening and perceptively increased enhancement commensurate with the increased tissue density ( Fig. 8.3 ). The mesentery demonstrates signal characteristics isointense to macroscopic fat without enhancement or fluid under normal circumstances. Cinegraphic motility images depict peristaltic activity that varies in pace at any given time, but maintains overall uniformity.

▪ Inflammatory Etiologies

The exquisite tissue contrast and multiparametric nature of MR enterography render it sensitive to small bowel inflammatory changes. However, availability and convenience considerations usually steer patients to CT in this setting. Inflammatory bowel disease (IBD), or Crohn’s disease, dominates the inflammatory category because of the need for repetitive surveillance imaging and the relatively young patient cohort.

▪ Crohn’s Disease

Crohn’s disease (CD) is a chronic inflammatory disease of the GI tract characterized by inflammatory exacerbations and regressions with disease onset usually in the second and third decades of life. Although idiopathic, evidence suggests an abnormal mucosal response to an unknown antigen. Chronic diarrhea is the most common presenting symptom and other symptoms include cramping abdominal pain, weight loss, low-grade fever, and anorexia.

CD threatens the entire GI tract (“mouth to anus”), but involves the small bowel (most commonly the terminal ileum [TI]) in approximately 80% of cases with colonic involvement in up to 50% of cases (usually with coexistent small bowel disease). Submucosal lymphoid hyperplasia and lymphedema develop first, reflected radiographically by mucosal aphthous ulcers. Shallow aphthous ulcers progress to deep then transmural ulceration, coalescing to the “cobblestone pattern.” Multiple noncontiguous (“skip”) segments of variable length usually feature asymmetric mural involvement associated with thickening of the surrounding mesentery.

The treatment strategy has evolved from focusing on symptom management and normalizing inflammatory biochemical markers to achieving full mucosal healing. As such, imaging surveillance and MR enterography play an important role in monitoring the response to treatment and guiding management. Distinguishing active inflammation—treated medically—from chronic fibrostenosing inflammation—treated surgically—is important.

An imaging-based classification scheme has been devised to standardize the assessment of inflammation and minimize subjectivity. The four part classification system includes: 1) active inflammatory, 2) perforating and fistulizing, 3) fibrostenotic, and 4) reparative and regenerative categories ( Table 8.2 ). Multiple studies have validated the high sensitivity (over 90%) of MR enterography for active inflammation in CD. MR signs of active inflammation include mucosal hyperenhancement (the most sensitive finding), prominence of the vasa recta (“comb sign”), stranding of the surrounding mesenteric fat and mural stratification ( Fig. 8.4 ). Stratification, or the “target sign,” arises from (T1)-hyperintense serosal hyperenhancement, hypointense submucosal edema, and hyperintense mucosal hyperenhancement. The appearance in T2-weighted images is essentially inverted. The overall thickness of the bowel wall ranges from 4 to 12 mm with occasional luminal stenosis. Identification of ulcers depends on luminal distention—the markedly T2-hyperintense, avidly enhancing mural defect surrounded by moderately T2-hyperintense, gradually enhancing mural edema is obscured by apposition of the adjacent bowel walls.

TABLE 8.2

Crohn’s Disease Classification System

Active Inflammatory		Fibrostenotic		Fistulizing/Perforating	Reparative/Regenerative
Minimal	Severe	Minimal	Severe
Superficial/aphthous ulcers	Deep ulcers/cobblestoning	Minimal luminal narrowing/mild prestenotic dilatation	Marked luminal narrowing/marked prestenotic dilatation	Deep fissuring ulcers and sinus tracts	Mucosal atrophy
Minimal fold thickening/distortion	Marked wall thickening/mural stratification	Minimal wall thickening	Marked wall thickening	Fistulas to adjacent bowel loops, skin	Regenerative polyps
	Mesenteric engorgement/ “comb sign”			Associated inflammatory phlegmon	Minimal luminal narrowing

Cinegraphic motility imaging helps to identify diseased bowel segments as conspicuously “frozen” against the background of normally peristalsing small bowel loops. Correlation of the “frozen bowel sign” with findings in the static pulse sequences—mucosal hyperenhancement, mural stratification, mesenteric inflammation, etc.—confirms active inflammation. Withholding antiperistaltic agents can make hypo- or aperistaltic diseased segments more conspicuous because normal bowel loops exhibit more active peristalsis.

With progressive disease, the transmural nature of CD leads to perforating disease in up to one third of patients. Fistulization extends either internally or externally—frequently occurring in the perineal region—and the reported sensitivity and specificity of MR ranges from 83.3%–84.4% and 100%, respectively. Nascent fistulas are linear, T2-hyperintense, peripherally avidly enhancing tracts; continuity of the enteric lumen with the fistulous tract clinches the diagnosis ( Fig. 8.5 ). Most fistulous tracts lack intraluminal contrast to confirm their presence and etiology. With progression, an internal fistulous tract incites a desmoplastic reaction in the surrounding mesentery, and with complication and involvement of at least two discontinuous segments, the imaging pattern often conforms to a stellate pattern, or the “star sign ( Fig. 8.6 ).” Extraintestinal complications include phlegmons and abscesses, mesenteric inflammation, and involvement of adjacent viscera.

Fibrostenotic disease is distinguished by the presence of bowel obstruction upstream from a fixed, narrowed segment of bowel. The fixed nature of the narrowing is reiterated in successive pulse sequences throughout the course of the examination and in the motility sequence ( Fig. 8.7 ). Without superimposed active inflammation, fibrostenotic bowel segments typically demonstrate mild, progressive enhancement and relative T2 hypointensity, reflecting fibrosis. Because of the asymmetric inflammatory involvement of the mesenteric border with subsequent asymmetric shortening of the diseased mesenteric-sided wall, pseudosacculations develop with relative ballooning of the antimesenteric bowel wall ( Fig. 8.8 ). Identifying these features indicating chronicity help to appropriately triage these patients to surgical resection of the diseased segment.

Reparative disease manifests with mucosal atrophy and regenerative polyps without inflammation or obstruction. Areas of mucosal denudation coexist with filiform polyps, which appear as punctate luminal filling defects.