Dr. Kelly and Dr. Perumpillichira are Clinical Fellows and Dr. Zalis is an Assistant Radiologist in the Division of Abdominal Imaging and Intervention, Department of Radiology, Massachusetts General Hospital, Boston, MA.

Computed tomography colonography (CTC) has received increasing interest since its first description in 1995 by Vining et al. ¹ This attention stems from the technique's potential to facilitate widespread, noninvasive colorectal carcinoma screening. This article will review the impetus for the technique and some of the supporting data, will cover the basic steps in performing the examination, and will briefly present the directions of current colonography research.

Colorectal carcinoma (CRC) is the second most common cause of cancer-related death in the United States and the developed world. Each year in the United States alone, approximately 55,000 individuals succumb to this disease and approximately 120,000 new cases are diagnosed. ^2,3 A great deal of investigation into the biology of CRC has led to the consensus that it commonly arises from a well-defined precursor lesion, the adenomatous polyp. ⁴ The current model of the adenoma-carcinoma sequence describes a process of degeneration from a small polyp to invasive cancer that may take up to 10 years. Polyp size correlates with the risk of harboring carcinoma; polyps 1 to 2 cm are estimated to have a 5% risk of harboring a malignancy, and those >2 cm have a >10% risk of malignancy. ⁵ Large, prospective, randomized, controlled studies have demonstrated that screening at-risk individuals for fecal occult blood and polyps can lead to a decrease of CRC-related mortality up to 30%. ⁶ As a result, a number of cancer societies, including the American Cancer Society, have advocated regular, complete structural evaluation of the colon to identify and subsequently remove polyps. ⁷

Since well before the development of CTC, several techniques have been used to evaluate the colon for CRC screening. Fecal occult blood testing (FOBT, often referred to as the guaiac test) and sigmoidoscopy are widely available and relatively inexpensive techniques; however, their performance is limited. In the case of FOBT, detection of colon lesions is compromised by limited sensitivity and a large number of false-positive results. ⁸ The performance of sigmoidoscopy is restricted by the fact that only approximately one-third of the colon is evaluated. ⁹ Radiologists are familiar with the barium enema (BE), long a mainstay of colon evaluation. Performance of BE for detection of polyps >= 1 cm has been estimated at 75% to 85%; however, this performance is believed to be dependent upon radiologist experience. ^10,11 Recently, referrals for BE have declined, due largely to the emergence of the current gold standard for colon screening: video-assisted colonoscopy. ¹² As a screening tool, colonoscopy has the advantage of being highly sensitive and simultaneously therapeutic--detected polyps can be removed in the same setting. However, the procedure entails a small but significant risk of complications, with serious morbidity or mortality occurring in approximately 0.2% of cases. ¹³ The impact of these complications is considerable if applied to the several million U.S. individuals considered at risk. In addition, colonoscopy is relatively expensive, in terms of direct costs for the procedure and the opportunity costs (eg, lost wages and other income) due to recovery from the sedation that the examination requires. ¹⁴

Despite the existence of colon screening methods, including colonoscopy, the mortality of CRC remains high, in part reflecting the poor compliance of at-risk individuals with recommended screening regimens. As an example, a recent U.S. government survey revealed that only 15% of Medicare beneficiaries undergo any colon cancer screening at all. ^15,16 If we focus on the gold standard of colonoscopy, given the expense, complications, limited access, and limited compliance for colonoscopy-based screening, clearly a need exists for a more easily tolerated, cost-effective method to perform colon screening.

In CT colonography, the abdomen and pelvis are scanned following a purging bowel preparation and gentle gas insufflation of the colon. The images, acquired in both prone and supine positions, are transferred to a computer workstation for coordinated multiplanar and three-dimensional (3D) evaluation of the reconstructed images. A radiologist then evaluates the CT model of the colon, looking for polyp lesions of the colon wall. CT colonography is limited to diagnosis only, and one may ask what the value of such a test is, if true-positive lesions detected on CTC require a subsequent colonoscopy for their removal. It is important to keep in mind that in a screening population, the prevalence of lesions, especially those likely to harbor carcinoma, is relatively low, typically in the range of 10% (lesions >= 1 cm). ^17,18 Hence, the value of CTC lies in its potential ability to correctly identify individuals with significant polyp lesions, while allowing the large majority of subjects to avoid the more involved and expensive endoscopic procedure.

Several studies have evaluated the performance of CTC compared with colonoscopy for the detection of polyps. ^19-24 As is often the case with developing techniques, these initial studies were performed using relatively small, polyp-enriched cohorts. The results of these studies can be broadly summarized as encouraging with qualification. The sensitivity of CTC for detection of polyps >= 1 cm--the lesions that are most likely to harbor carcinoma--ranged from approximately 85% to 93%, with a trend toward higher detection rates in more recent studies. Improved performance can be attributed to improved scanning technology (leading to decreased motion and volume-averaging artifacts), improved workstation software, and improved radiologists' experience with the technique. Specificity for detecting polyps >= 1 cm (usually defined on a per-segment basis) ranged from 90% to 98%. For lesions <1 cm in size, the performance of CTC decreases, with sensitivities and specificities estimated at 66% to 82% and 63% to 97%, respectively, for lesions 5 to 9 mm. While the results of CTC are encouraging, performance in these studies may have been influenced by the relatively high prevalence of polyps in the enriched cohorts examined, hence final validation of CTC as a primary colon screening examination awaits the results of the large, screening-cohort, multicenter trials now being organized.

Nonetheless, in addition to encouraging performance statistics, two very important patterns emerge from the data currently available. First, CTC is a safe examination. To date, there have been no reports of perforation or other serious complications in more than 1000 cases reported in the literature. In addition, patients seem to prefer CTC over colonoscopy. Svensson et al ²⁵ reported that among patients receiving both colonoscopy and CTC, the latter is preferred by the majority of patients who demonstrated a preference. Furthermore, 94% subjects in this study described CTC as not difficult or slightly difficult.

Based on these data and accumulating experience, it is reasonable to discuss two thresholds of performance for CTC. The first--the use and promotion of CTC as the primary method for colon cancer screening--has not yet been validated in a properly designed clinical screening trial. These trials are now recruiting subjects, and it is hoped that their results will emerge within the next several years. The second threshold--the use of CTC as a safe and accurate alternative to colonoscopy for detection of polyps--is supported by the studies published to date. In this application, CTC is intended for patients either unwilling or unable to undergo endoscopy, as an attempt to strike a balance between the evident need for a more easily tolerated colon examination and the limited, but encouraging, data now available.

In addition, CTC has proven useful in cases of incomplete colonoscopy. When the endoscopist cannot reach the cecum, for example due to tortuosity of the colon, an obstructing lesion, or patient discomfort, CTC can complete the colon examination, often the same day as the failed colonoscopy (Figure 1). ^26,27 As a result, several centers now perform so-called completion colonography on a routine basis.

Balancing performance statistics, immediate and downstream costs, and patient outcomes with CTC as a screening modality merits a thorough decision analysis, which to our knowledge has yet to be performed.

Performing the basic CTC examination

The examination process should begin with patient education before the image acquisition. Patients should be reassured that CTC is usually a very easily tolerated examination, and that it is generally completed within 20 minutes. No intravenous (IV) contrast is required for the vast majority of screening patients. Patients can return to their regular diet and activity immediately after the examiniation; there is no recovery period, per se. While it may be possible in some centers to perform online reading immediately after images acquisition, this requires close coordination and flexibility on the part of the radiology and endoscopy units, and likely will remain impractical in most facilities. Hence, most patients should expect their results to be forthcoming, reported by the radiologist to their referring physician. Patients must understand that CTC is a screening and diagnostic test only; in the event that a significant abnormality is detected in the colon, they may be referred for colonoscopy for polypectomy or biopsy. Finally, it is essential to emphasize the importance of the pre-examination purging bowel preparation for the performance of a high-quality examination.

While no clinical trials have demonstrated significant differences in polyp detection rates with the various bowel preparation kits, it is useful to understand some of the differences involved with them. Two basic kinds of preparation are currently in use, a so-called dry prep (based on a combination of osmotic and pharmacologic cathartics, such as bisacodyl sodium, phospho-soda, and magnesium citrate), and a wet prep (based on nonabsorbable orally ingested polyethylene glycol electrolyte [PEG] solution). Each type of preparation has advantages and disadvantages. The dry preps involve smaller quantities of substance to be ingested (small volume pills, powders, or suppositories) and usually result in less retained fluid within the colon. ²⁴ The last point is relevant because polyps can be obscured if they are submerged in water density fluid, and their evaluation can be made more complicated if they are visible on only one of the prone or supine views. However, the dry agents can cause cramping in some individuals, and the salt load associated with some of these preparations may be of concern for individuals with renal or congestive heart failure. In addition, if compliance with the dry prep is incomplete, these agents can result in residual desiccated fecal material adhering to the colon. The desiccated material can appear nearly identical to polyp lesions and can result in multiple pseudolesions that complicate the radiologist's evaluation. By contrast, the wet prep requires the patient to ingest a relatively large volume of nonabsorbable fluid the night before the examination. The wet prep is safe to use in patients with compromised renal or cardiac function. In addition, little desiccated material is left behind, in part because of the fluid ingested with the prep. While more fluid may be present in the colon, no studies have demonstrated a statistically significantly decrease in polyp detection rates with the wet prep.

Once a patient has taken the preparation and reported for the examination, the patient must undergo gas insufflation of the colon. Insufflation can be accomplished using either carbon dioxide (CO ₂ ) or room air, and is easily tolerated by the majority of patients. While it has been suggested that CO ₂ would be more easily tolerated by patients because of its more rapid reabsorption from the bowel, initial evaluation by McDermott et al ²⁸ demonstrated no significant difference in observed distension or reported patient comfort when they compared room air with CO ₂ . As first reported by Yee, ²⁹ and confirmed by growing consensus, most investigators no longer use spasmolytic agents for colonography insufflation. ^30,31

Administration of the gas (air or CO ₂ ) can be performed either by manual pumping of an insufflator bulb attached to a small, flexible rectal tube, or semi-automatically, by means of a pressure-regulated mechanical insufflator. In our institution, we use a 16F Foley catheter attached to a hand bulb insufflator. In most cases, we obtain adequate insufflation with 30 to 50 puffs of air, as tolerated by the patient. Hand bulb insufflation is less expensive in terms of equipment, but requires 1 to 2 minutes of attention by the CT staff. Automated insufflators are conceptually attractive in that they can gently achieve adequate distension and maintain it during the procedure. However, automated insufflators can be problematic in the setting of a stricture, partial obstruction, or bowel cramping, wherein the pressure regulation system--which typically senses pressures only in the distal colon--can prematurely terminate the insufflation, leaving the right colon relatively collapsed. Anecdotally, several investigators have reported successful use of automated insufflators, but to our knowledge, formal comparison of automated with manual technique has not yet been published. ²⁸

Regardless of which combination of gas and insufflator system is used, it is important to realize that the insufflation process need not occur rapidly--gentle administration of gas over 1 to 2 minutes greatly reduces the likelihood of bowel cramping, and improves patients' perception of the examination. Some investigators have advocated self-insufflation of the colon, and pilot data suggest that motivated patients can adequately control the rate and volume of gas by themselves, once the radiology staff has inserted the rectal tube. ³²

Image acquisition for CTC should be performed on a helical CT scanner, preferably with at least a 4-row multidetector scanner. Multidetector technology greatly reduces the scan time, permitting single breath-hold acquisitions, and minimizes motion artifact. Scan technique can exploit the naturally high contrast between the gas of the insufflated bowel lumen and the soft-tissue density of colonic polyps. As a result, a markedly reduced X-ray technique can be used, typically in the range of 40 to 100 mAs, with 100 to 140 kVp. These scan parameters result in a total radiation dose to the patient that is somewhat less than that of BE. ²²

Images should be acquired with ¾ 5 mm collimation and reconstructed with 50% overlap, to facilitate adequate multiplanar and 3D reconstruction. For example, on a GE Lightspeed scanner (GE Medical Systems, Waukesha, WA) using the HS or fast table speed mode, an adequate protocol consists of 50 mA, 140 kVp, 3.75-mm collimation, with slices reconstructed at 1.8-mm intervals. Patients should be scanned in both prone and supine positions, in order to maximize visualization and distension of all regions of the colon.

With improving scanner technology, it will be feasible to image with isotropic or near-isotropic voxels, provided these techniques do not adversely impact patients' absorbed radiation dose or PACS logistics.

The use of IV contrast for CTC examinations is currently under investigation. For screening examinations, addition of IV contrast adds to the cost of the examination and exposes patients to a small but demonstrable risk of contrast reaction. Morrin et al ^33,34 have reported increased lesion conspicuity and reader confidence with IV contrast, in particular for lesions submerged in residual bowel fluid. The added benefit of IV contrast in the evaluation of a screening population has not yet been evaluated. However, when a known tumor is present, the addition of IV contrast permits evaluation of tumor invasion and metastatic disease. In the setting of incomplete colonoscopy secondary to colonic mass, CTC can simultaneously evaluate the proximal colon, as well as diagnose and stage with one imaging study. ^26,35,36

Colonography interpretation

Radiologists should evaluate CTC on soft-copy workstations only. It is not feasible to evaluate the large number of images in printed form, and real-time, multiplanar interaction with the data is essential for confident evaluation. The popular synonym for CTC, virtual colonoscopy, is in part derived from the initial use of 3D endoluminal reconstructions for evaluation of the image data. Most investigators now advocate endoluminal evaluation only as a problem-solving tool, as described below. ^21,22

Polyps are invariably uniform soft-tissue lesions that arise from fixed positions along the colon mucosa. As such, they characteristically demonstrate a homogeneous soft-tissue density in cross section, and except for the relatively rare pedunculated lesion, they do not move on prone and supine images (Figures 2 and 3). Mucosal objects that contain bubbles of air or that move with changing patient position are distinguishable as retained feces. In addition, polyps have a characteristic morphology; in the majority of cases they present as mushroom-capped or sessile lesions. Infiltrating, sessile lesions may be subtle on CTC. Irregular thickening of a haustral fold should raise suspicion of an underlying lesion. While there has been concern that CTC would not be able to detect these lesions, Fidler et al ³⁷ have reported a sensitivity of 89% for flat lesions >=1 cm. Importantly, polyps are generally not linear in morphology.

The dynamic tracking of the axial data with judicious use of multiplanar or endoluminal views can permit a time-efficient evaluation of observed candidate lesions, and, with radiologist experience, permit a highly accurate evaluation. ²¹ With current software, an experienced reader can expect to complete the evaluation of the colon and the other soft tissues of the abdomen and pelvis in <15 minutes.

The radiologist interpreting a CTC should report the extracolonic soft-tissue findings as well as those in the colon. Hara et al, ³⁸ evaluating a cohort of 264 patients who underwent CTC, found that 30 (11%) had significant extracolonic findings. Eighteen of these 30 patients underwent additional diagnostic imaging, and 5 ultimately underwent surgery for the conditions--both malignant and nonmalignant--initially observed on CTC. As extracolonic findings may trigger further work-up and often entail cost and anxiety for the patient, it is important that they are described with an appropriate estimation of their clinical significance. As collective experience with CTC increases, we can hope to establish more refined guidelines for the reporting of indeterminate extracolonic findings.

False-positive results are most often related to retained stool or bulbous folds. Careful examination of both prone and supine images, as well as evaluation of the 3D views, are necessary to limit this potential pitfall. Stool is characteristically heterogeneous, containing low attenuation foci of air, and shifts position between prone and supine images (Figure 4). Although villous lesions occasionally trap air between the fronds of tumor, there is almost invariably a solid soft-tissue density component. Redundant folds are smooth in contour, and will often distend with change in patient position. The ileocecal valve may also be mistaken for a polypoid lesion, but with experience, a radiologist can identify it confidently by the combination of its characteristic location, dimpled shape, and (frequently) fatty content.

Limitations and future directions

Computer-aided diagnosis (CAD) is one of the methods under investigation to further decrease interpretation times and increase accuracy. Ideally, the interpretation time for CTC could be cut from 15 to 5 minutes, and CAD could potentially augment reader performance. Current algorithms exploit characteristic changes in mucosal curvature and soft-tissue density that occur in polyps in order to identify potential polyp candidates. Two published reports in limited, enriched cohorts are highly encouraging that computer-based techniques may assist the radiologist. ^39,40 One challenge arising in this avenue of research is the need to balance sufficient sensitivity for polyp detection while limiting the number of false-positive calls by the computer. ⁴¹

Another important area of research in CTC focuses on the pre-examination bowel preparation. Bowel cleansing is the most unpalatable part of any colon examination for many patients and the need for bowel preparation has been noted as a reason for low patient compliance with recommended screening strategies. ^42,43 Modifications of the CTC protocol that reduce or eliminate bowel preparation are under investigation. In these techniques, the patient ingests a positive contrast agent, such as barium or iodinated contrast, with meals for 1 to 2 days prior to the examination. Contrast admixes with ingested food and becomes homogeneously incorporated into stool, rendering it distinctly high in attenuation on CT. Callstrom et al ⁴⁴ and Lefere and coworkers ⁴⁵ reported thorough tagging of stool using barium contrast. Zalis et al ⁴⁶ further reported the use of software to electronically remove the opacified stool, a process called digital subtraction bowel cleansing, which retains the utility of 3D views in submerged lesions (Figure 5). Initial results of these approaches are encouraging, demonstrating an 85% to 89% sensitivity in small series. ^44,47

Conclusion

There has been rapid, exciting progress in the development and implementation of CTC for detailed evaluation of the colon. CT colonography offers the potential of a more easily tolerated, noninvasive method to perform colon cancer screening, one that may ultimately be implemented with computer-assisted reading and without use of a purging bowel preparation. Currently, the examination has proven useful as an alternative detection method when complete endoscopic evaluation is not possible. The role of CTC in widespread colon cancer screening will be clarified as its performance in larger screening cohorts is evaluated thoroughly. AR