VIRTUAL COLONOGRAPHY

Virtual colonography (VC) is based on either magnetic resonance (MR) or computed tomography (CT) 3D datasets. Based on cross sectional images, this type of examination offers several advantages over conventional colonoscopy, the most significant of which relates to the lack of procedural pain and discomfort. Angtuaco et al demonstrated that more than 60% of potential patients preferred VC over conventional colonoscopy when both methods were offered.¹³ Non-invasiveness and no need for sedation were the main reasons for the preference of virtual colonography.

In contrast with a conventional colonoscopic analysis, VC is not limited to endoscopic viewing. Rather the 3D datasets can be scrolled in a traditional 2D mode on a workstation in any desired plane. This type of multiplanar reformation analysis depicts the colonic lumen as well as the colonic wall relative to the surrounding abdominal morphology. Lesions can thus be located more accurately. Even in the presence of stenotic tumours, the entire colon can be assessed, which often is not possible in conventional endoscopy. A further advantage of VC based on either CT or MR datasets relates to the simultaneous assessment of other abdominal organs. In the case of patients with suspected colorectal tumour, imaging of the liver must be considered a welcome addition.

Despite the availability of multiple studies reporting its high diagnostic accuracy,^14,15 the future of CT colonography as a screening method for colorectal cancer is uncertain: patients are exposed to considerable doses of ionising radiation.^16,17 This is compounded by the fact that two datasets, one in the prone and a second one in the supine position, need to be collected. As a screening examination of the colon should be repeated at regular intervals (every three to five years), the radiation issue may even evolve into a public health concern. Therefore, it seems reasonable to focus on MRI for colorectal screening strategies. The technique is not associated with any exposure to ionising radiation and lacks any other known harmful side effects. Furthermore, contrast agents applied in conjunction with MRI examinations are characterised by a more favourable safety profile than CT contrast agents as they lack any nephrotoxicity and are associated with far fewer anaphylactic reactions.^18,19

VIRTUAL MR COLONOGRAPHY: TECHNIQUE

Currently two techniques are being evaluated for MR colonography (MRC). Based on the signal within the colonic lumen, they can be differentiated as “bright lumen” and “dark lumen” MRC.

Bright lumen MRC

Similar to contrast enhanced 3D MR angiography, MRC is based on the principles of ultra fast, T1 weighted 3D GRE acquisitions collected within the confines of a single breath hold.²⁰ This requires the use of an MR scanner equipped with high performance gradients. To permit homogenous signal transmission and reception over the entire colon with high CNR values, a combination of phased array surface coils should be used. The size of the coil must permit coverage of the entire colon.

As colonic lesions often cannot be differentiated from stool, the patient has to undergo bowel cleansing in a manner similar to that required for conventional colonoscopy. Before the examination the patient should be screened for contraindications to MRI such as severe claustrophobia, presence of metallic implants in critical regions such as the eyes, spinal chord or brain, or cardiac pacemakers. The presence of hip prostheses, which normally is not regarded a contraindication to MRI, impedes a complete analysis of the rectum and sigmoid colon. Therefore, patients with hip prosthesis should also not be examined by MRC.

After placement of a rectal enema tube, the colon is filled with the patient in the prone position using 1000 to 2000 ml of a water based enema, spiked with paramagnetic contrast (1:100). The enema is administered using 100 cm–150 cm of hydrostatic pressure (fig 1). To reduce bowel motion and alleviate colonic spasm, the use of intravenously administered spasmolytic agents (for example, scopolamine or glucagon) before and during the bowel filling is helpful. In contrast with conventional colonoscopy sedative or analgesic agents do not have to be applied. To ensure safe and optimal bowel filling and distension, the filling process is monitored with a non-slice select 2D acquisition, collecting one image every three seconds (fig 2). Once the enema has reached the caecum, a 3D dataset of the abdomen encompassing the entire colon is collected. To compensate for the presence of residual air exhibiting “filling defects” similar to polyps within the colonic lumen, 3D datasets are collected in both the prone and supine patient positions. Hereafter the enema bag is placed on the floor for facilitated emptying of the colon and the patient is removed from the scanner.

The acquired 3D MR datasets consist of coronal sections, ranging in thickness between 1.5 mm and 2 mm. The sequence is based on the use of short repetition (TR 1.6 ms–3.8 ms) and echo times (0.6 ms–1.6 ms). The achievable minimum TR should be shorter than 5 ms; otherwise, the acquisition of a 3D dataset cannot be collected within the confines of a single breathhold. In conjunction with a field of view of 400×400 mm and an imaging matrix of 460×512, the spatial resolution includes an interpolated voxel size of about 1 mm×1 mm×1.6 mm.

On the 3D GRE datasets only the colonic lumen containing the enema is bright whereas all other tissues remain low in signal intensity. The resulting contrast between the colonic lumen and surrounding structures is the basis for subsequent virtual colonographic viewing (fig 3). The MRC protocol can be further amplified by the acquisition of 2D gradient echo datasets after the intravenous application of a gadolinium containing contrast compound. This permits a more comprehensive assessment of parenchymal abdominal organs and increases the ability to detect hepatic metastases.

Bright lumen MRC can be completed within 20 minutes, including the time for patient positioning, image planning, and data acquisition. The 3D datasets are subsequently processed using commercially available software and hardware. A complete analysis of an MRC examination still requires 15 minutes of interactive image viewing on a high performance work station. In a first step MRC images should be interpreted in the multiplanar reformation mode scrolling through the prone 3D dataset in all three orthogonal planes. In regions containing larger pockets of residual air, the assessment needs to be supplemented by views of the supine dataset. In a second step the data should be assessed based on virtual endoscopic renderings displaying the inside of the colonic lumen. A virtual endoscopic fly through allows the observer to concentrate on the colon facilitating the depiction of small structures protruding into the colonic lumen. Furthermore, the three dimensional depth perception permits the assessment of haustral fold morphology, thereby increasing the observer’s ability to distinguish polyps from haustra. To assure complete visualisation of both sides of haustral folds, the virtual fly through should be performed in an antegrade as well as retrograde direction (fig 4).

Dark lumen MRC

The detection of colorectal lesions with “bright lumen” MRC relies on the visualisation of filling defects. Differential considerations for such a filling defect beyond polyps include air bubbles as well as residual fecal material. To permit differentiation datasets are collected in both the prone and supine patient position: air and fecal material move, while polyps remain stationary. While effective in most instances, the technique can introduce errors. Thus, polyps with a long stalk may move sufficiently to impress as a moving air bubble or more probably residual stool, while stool adherent to the colonic wall may not move at all and thus falsely impress as a polyp. In addition to obviating the need for a second, time consuming 3D data acquisition “dark lumen” MRC facilitates the identification of polyps.

“Dark lumen” MRC focuses on the colonic wall. It is based on contrast generated between a brightly enhancing colonic wall and a homogeneously dark colonic lumen.²¹ The technique differs from “bright lumen” MRC in the following manner:

1. Instead of agadolinium containing enema only tap water is rectally applied rendering low signal on heavily T1 weighted 3D GRE acquisitions.

2. The colonic filling process is monitored with a fluoroscopic T2w sequence, rather than a T1w sequence.

3. To obtain a bright colonic wall paramagnetic contrast is applied intravenously. 3D datasets are collected before the application and after a 75 second delay.

4. As residual air exhibits no signal in the colonic lumen, the examination needs to be performed only in the prone patient position.

Compared with “bright lumen” MRC that has been extensively evaluated in the past, “dark lumen” MRC harbours considerable advantages including reduced examination and post-processing times, as only one 3D dataset needs to be collected. Furthermore, the “dark lumen” technique copes with the problem of residual stool in a simple manner: if the lesion enhances it is a polyp, if it does not enhance it represents stool (figs 5, 6). Suspicious appearing lesions are analysed by comparing signal intensities on the pre-contrast and post-contrast images. If analysis were limited to the post-contrast dataset, bright stool could be misinterpreted as a polyp. Comparison with the pre-contrast images records the lack of contrast enhancement, which assures the correct diagnosis.

Enhancement of colorectal masses following the intravenous administration of contrast has been reported in conjunction with MRC²² and CT colonography.²³ The use of intravenously administered contrast material significantly improves reader confidence in the assessment of bowel wall conspicuity and the ability to depict medium sized polyps in suboptimally prepared colons. The enhancement observed within polyps exceeds the increase determined within the colonic wall. This may aid in differentiating even very small polyps from thickened haustral folds.

A further advantage of “dark lumen” MRC relates to the fact that it permits direct analysis of the bowel wall. This might facilitate the evaluation of inflammatory changes in patients with inflammatory bowel disease (fig 7). Increased contrast uptake and bowel wall thickening, as recorded on contrast enhanced T1 weighted images has already been shown to correlate well with the degree of inflammation in the small bowel.²⁴ Hence, the “dark lumen” approach may indeed amplify the list of indications for MRC in the future to also encompass inflammatory bowel disease.

Finally, the intravenous application of paramagnetic contrast permits a more comprehensive assessment of parenchymal abdominal organs contained within the field of view. By combining pre-contrast and post-contrast T1 weighted imaging, the liver can be accurately evaluated regarding the presence and type of concomitant disease. Dark lumen MRC also offers new perspectives regarding optimisation of bowel distension. Although the administration of water as a rectal enema does not adversely affect patient comfort in most cases, a modified strategy could be based on the application of gases like carbon dioxide.²⁵ The gas is signalless and would thus easily permit delineation of the contrast enhanced colonic wall and masses.

DIAGNOSTIC PERFORMANCE OF MRC

The diagnostic performance of bright lumen MRC was assessed in several studies^26,27 using conventional colonoscopy as the standard of reference. While most mass lesions smaller than 5 mm in size were missed,¹³ almost all lesions exceeding 10 mm were correctly identified (table 1). In a study by Pappalardo et al²⁸ MRC even detected a higher total number of polyps exceeding 10 mm in size than conventional colonoscopy. MRC identified additional polyps in regions of the colon not reached by colonoscopy.

Direct observational data on growth rates indicated that polyps smaller than 10 mm remain stable over three years and are not prone to malignant degeneration.²⁹ Hence, bright lumen MRC may be considered almost as reliable as conventional colonoscopy regarding the assessment of colonic lesions at risk for malignant degeneration. Nevertheless, attempts are under way to increase the spatial resolution of the underlying 3D datasets and thereby improve the diagnostic accuracy of MRC for lesions ranging between 5 mm and 10 mm in size. Technique refinements include the use of even shorter repetition times in conjunction with zero filling routines and the implementation of parallel imaging routines.³⁰

Fecal tagging

MRC still requires bowel cleansing in a manner similar to conventional colonoscopy. As 75% of patients undergoing bowel preparation complain about symptoms ranging from “feeling unwell” to “inability to sleep”,³¹ patient acceptance is affected negatively. To assure high patient acceptance of MRC, bowel cleansing needs to be eliminated. This can be accomplished with fecal tagging—a concept based on modulating the signal intensity of fecal material by adding contrast compounds to regular meals.

Fitting the two approaches to MRC (bright lumen and dark lumen), there are also two theoretical approaches to fecal tagging. Its principle was demonstrated on the basis of a bright rectal enema distending the colonic lumen containing brightly tagged stool in conjunction with bright lumen MRC.³² By adding a T1 shortening Gd based MR contrast agent to regular meals before the MR examination, harmonisation of signal properties between fecal material and the Gd based enema was achieved. The oral administration of a paramagnetic MR contrast agent (Gd-DOTA) has been shown to be safe. The combination of fecal tagging with a paramagnetic contrast agent and colonic filling results in a homogenous signal distribution throughout the colon (fig 7). In these examinations virtual MRC permits an unobstructed view through the colon because the tagged stool is virtually indistinguishable from the administered enema. Although encouraging results concerning acceptance and image interpretation were obtained, the clinical implementation of bright lumen fecal tagging was hindered by the high cost of the Gd based paramagnetic contrast agent.

A second strategy for fecal tagging is based on rendering the colonic lumen dark.³³ For fecal tagging, a highly concentrated, barium sulphate containing contrast agent (Micropaque; Guerbet, Sulzbach, Germany; 1 g barium sulphate/ml) is administered in a volume of 200 ml with each of four main meals beginning 36 hours before MRC. Patients are instructed to avoid the intake of all fibre rich foodstuff and nourishments with high concentration of manganese such as chocolate or fruits during this period, as manganese leads to an increased signal intensity in T1w sequences. “Barium based” fecal tagging is combined with dark lumen MRC: the colon is distended with a rectally applied water enema and paramagnetic contrast is administered intravenously to render the colonic wall and adherent colorectal mass lesions bright.

Barium sulphate is a well known diagnostic contrast agent, still in common use as an oral agent for oesophageal, gastric, and small bowel radiography. Compared with Gd based contrast compounds, it is far less costly and characterised by an even better safety profile. Anaphylactoid reactions or other adverse side effects are virtually unknown. The agent is not absorbed and mixes well with stool. Thus, barium includes all characteristics as an ideal oral tagging agent for MRC.

The barium based approach to fecal tagging has been successfully assessed.³³ The signal reducing effects upon stool has been documented in volunteer studies. By ingesting barium before the MR examination, stool is rendered virtually indistinguishable from the administered water enema on heavily T1w 3D GRE images (fig 8). The MR examination without prior ingestion of barium reveals signal rich stool that cannot readily be differentiated from the brightly enhancing colonic wall (fig 8).

Recently, the barium based fecal tagging concept has been successfully evaluated in a pilot patient study. Fecal tagged MRC detected all polyps larger than 8 mm in a population of 24 patients with known or suspected colorectal tumours.³⁴ Overall sensitivity of MRC amounted to 89.3% for the detection of colorectal masses, and specificity was 100%. Colorectal cancers (fig 9) and polyps (fig 10) were readily identified as such.

Although further work is required to confirm these excellent results, it seems that barium tagged MRC has vast potential to emerge as the examination strategy of choice for the early detection of polyps in asymtomatic subjects. The technique seems to combine excellent diagnostic accuracy with high patient acceptance based on a painless examination and no need for colonic cleansing.