Applied Radiology

Dr. Raziano is currently a fourth-year Resident in Radiology at Emory University Hospital, Atlanta, GA. He received his MD degree from Tulane University School of Medicine, New Orleans, LA in 1999. He will remain at Emory for a Neuroradiology fellowship following completion of his residency program. Dr. Torres is a Professor of Radiology, Division of Abdominal Radiology, Emory University Hospital, Atlanta, GA.

The effectiveness of screening and polypectomy in reducing colon cancer incidence and mortality is well known. Virtual colonoscopy, an advanced application of computed tomography (CT), has demonstrated great potential for revolutionizing colon cancer screening. Multislice helical CT technology, three-dimensional (3D) reconstruction techniques, and lower radiation dose protocols have significantly improved the sensitivity of virtual colonoscopy. In addition to screening, several other important applications of CT colonography are under investigation. One important new application is the evaluation of patients in whom tissue diagnosis has already been obtained. Appropriate staging of such patients is essential for avoiding unnecessary morbidity and maximizing therapeutic benefit. Virtual colonoscopy plays an important role in preoperative evaluation by accomplishing both surveillance for metastases and evaluation of mucosal disease in a single procedure while minimizing patient discomfort. CT colonography is helpful in postoperative surveillance as well. Intravenous contrast plays an important role in these pre- and postsurgical patients by improving both visualization and characterization of lesions and overcoming many of the previous limitations of virtual colonoscopy.

Colorectal cancer is the third most common type of cancer and the third leading cause of cancer death in the United States. The National Cancer Institute estimated that in 2002 there were 148,300 new cases of, and more than 56,600 deaths from, colorectal cancer. ¹ The incidence of colon cancer in the general population is about 0.3%, whereas it is about 15% for those with a family history of the disease. The majority of patients, approximately 75% of those who develop colon cancer, have no particular risk factors, an observation that emphasizes the need for broad-based screening. ¹

The pathogenesis of colon cancer is well established. It progresses slowly, over several years' time, from adenomatous polyp to invasive carcinoma, through the accumulation of genetic defects. ² This protracted course is fundamental in the selection of colon cancer as a potentially preventable disease. In order for screening to be effective, there must be a relatively long preclinical period during which detection of disease can have a substantial impact on patient outcome. The size of a colorectal polyp is directly correlated with the likelihood of malignant foci within, and it can take up to a decade for a cancerous lesion to develop. ^3-5 Thus, in the asymptomatic patient, screening and polypectomy can reduce mortality by prompting therapy in the early stages of disease. Resection of precancerous polyps has been convincingly demonstrated to reduce the overall incidence of colon carcinoma. ^6-8

Current screening protocols

Current recommendations of the American Cancer Society call for one of the following screening options beginning at age 50: yearly fecal occult blood test, alone or in combination with flexible sigmoidoscopy every 5 years; double-contrast barium enema every 5 years; or colonoscopy every 10 years.

The most widely used method, fecal occult blood testing, has a very low sensitivity (10% to 15%). ⁹ False positives can occur for a variety of reasons, a characteristic that further limits the value of this technique. Common causes include a recent meal of red meat, hemorrhoids, iron supplementation, mineral oil laxatives, and menstrual bleeding. Flexible sigmoidoscopy, another widely used screening technique, examines only the distal colon, leaving much of the colon unsurveyed. ¹⁰ Studies have demonstrated not only the need for more proximal screening, but also the need for evaluation of the ascending and transverse colon in those with positive findings on sigmoidoscopy. Furthermore, it has been demonstrated that approximately 50% of malignancies would be overlooked if only distal surveillance were performed. ¹¹

Barium enema has long been a mainstay of colonic evaluation. This method, however, has many limitations. Patients undergoing this procedure must be able to position themselves appropriately, so that contrast can traverse the colon. Redundancy of the colon, often encountered in the elderly, can frequently cause an incomplete examination. Moderate sensitivity for clinically significant polyps (75% to 85%), combined with significant radiation exposure and patient discomfort, necessitate a more benign alternative. ¹²

The current standard for evaluation of the colonic mucosa is direct visualization via colonoscopy. However, this is impractical in a number of circumstances. Limited resources prohibit the use of colonoscopy in every case in which it would be appropriate. Many patients are unwilling or unable to undergo colonoscopy, due to the discomfort either associated with bowel preparation or attributable to the procedure itself. Conventional colonoscopy is incomplete in as many as 5% of cases, where redundant colon prohibits passage of the colonoscope. ¹³ Moreover, despite colonoscopy's current status as the gold standard, up to 27% of polyps measuring <5 mm go undetected, according to a recent study in which initial colonoscopy was followed up immediately by a repeat procedure. ¹⁴

Patient noncompliance is a major factor further limiting the effectiveness of screening modalities. Fewer than 25% of Americans over the age of 50 follow the American Cancer Society guidelines for screening. This has largely been attributed to the discomfort associated with the aforementioned methods. ¹⁵ Despite the poor rate of compliance with suggested screening protocols, colorectal cancer death rates have decreased by 1.8% per year from 1992 to 1998, a trend largely credited to earlier detection and aggressive therapy. ¹⁶ It is clear from Figure 1 that co-lorectal cancer in the general population is most often a local or regional disease, amenable to curative resection. Clearly, then, screening can be of greater benefit if it is more widely accepted, as there is a high proportion of lower grade, potentially curable cancers.

CT colonoscopy, also called virtual colonoscopy, is currently being developed to address such issues. Patient tolerance for CT colonoscopy has been shown to be significantly better than for conventional colonoscopy, an observation that offers hope for more widespread screening. In one recent survey of 104 respondents, 68 patients favored one examination; of these, 56 (82%) preferred CT colonography over conventional colonoscopy, citing pain as one of their foremost concerns. In fact, nearly twice the number of patients in this study (49) felt that CT colonoscopy was not painful, compared with 28 in the conventional colonoscopy group. ¹⁵

The ability of this technique to detect clinically significant polyps is well documented and its utility in difficult cases confirmed. ^17,18 In a recent study of 100 patients published in the New England Journal of Medicine , CT colonoscopy had a sensitivity of 82% for all polyps and a specificity of 84%. More importantly, however, for larger (ie, clinically significant) polyps the sensitivity was 91%. ¹⁹ The applicability of this method to difficult cases was recently confirmed; 40 patients in whom complete colonoscopy could not be performed were given a virtual colonoscopic examination. In >90% of cases, the previously obscured colonic segment could be adequately evaluated using virtual colonoscopic techniques. ²⁰

In addition to its role in screening, virtual colonoscopy has great potential in the evaluation of patients with known or probable colorectal cancer. The survival rate in such patients is directly related to the extent of disease at the time of diagnosis. Those with local disease have a survival rate of 90% to 100%, whereas survival drops precipitously when distant metastases are present (Figure 2). ¹ Thus, early diagnosis is critical and offers the best chance for survival.

Treatment of colorectal malignancy depends upon its stage at the time of diagnosis. Local disease isolated to the mucosa requires surgical resection, often without the need for adjuvant chemotherapy. More invasive disease, characterized by varying involvement of the muscular wall of the colon, with or without local nodal disease, can be treated with a combination of surgery and chemotherapy. Disease that is metastatic at the time of discovery is generally treated with chemotherapy, with or without debulking surgery. When hepatic metastases are present, lesion resection or ablation is sometimes also possible, depending upon the extent of disease. Radiation therapy can be used to varying degrees, depending upon the size and extent of the malignancy; it is critical in particular for this intervention that the location and extent of disease be well established.

Virtual colonoscopy is ideal for cancer staging, as it allows for the evaluation of mucosal lesions, which are poorly demonstrated by standard CT. Given that approximately 7% of patients have additional lesions, two-thirds of which are synchronous and the remainder metachronous, it is essential that these patients be thoroughly evaluated prior to surgical therapy. ^11,21 Virtual colonoscopy combines the advantages of colonoscopy and CT staging, permitting simultaneous evaluation of mucosal disease as well as regional and distant metastatic disease. ²²

Local and nodal extension

Extension of tumor first through the mucosa and then into the muscular layers of the colon and beyond can be subtle, particularly in the collapsed bowel. The addition of pneumocolon aids in identification of aberrant wall thickening, and intravenous (IV) contrast further increases the conspicuity of the lesion. Contrast also plays an important role in the identification of nodal disease. Evaluation of local and nodal disease permits surgical planning that maximizes the chance of curative resection and may determine the need for chemotherapy. A study by Harvey et al ²³ demonstrated an overall staging accuracy approaching 80% with contrast-enhanced CT colonography, in much earlier stages of its development.

Hepatic metastases

The liver is typically the first organ to demonstrate metastatic involvement, making it a critical organ to evaluate in the determination of disease extent. Contrast-enhanced imaging significantly increases sensitivity for detecting hepatic metastases. ²⁴ Furthermore, patterns of enhancement are helpful in differentiating malignancy from benign hepatic lesions such as hemangioma. Characterization of the degree of hepatic involvement also allows for therapeutic intervention, if lesions are appropriately localized. Multidetector CT now permits accurate characterization of the vascular anatomy of the liver, which is critical to appropriate surgical planning. Not only does this affect decisions about resectability of the patient, but also allows for critical identification of anatomic variants. A recent paper by Sahani et al ²⁵ demonstrates the utility of hepatic CT angiography; 42 patients who were to undergo hepatic resection underwent CT angiography; in 17 of these patients, arterial anomalies were detected. This compared very favorably with the catheter angiography performed in 22 of these patients (accuracy, 97%; sensitivity, 94%; specificity, 100%), and was of significant benefit in surgical planning. ^25,26

Other metastases

In addition to the liver, other common sites of metastatic disease from colon cancer are the lung, adrenal gland, bone, kidney, pancreas, spleen, and central nervous system. Since the vast majority of patients with distant metastatic disease first have involvement of the liver, it is important to begin by identifying any hepatic lesions suggestive of metastatic disease. CT colonoscopy, therefore, plays a critical role in the appropriate staging of patients, permitting evaluation of mucosal, local, and metastatic disease in a single procedure.

Beyond staging: Surveillance for local recurrence

Approximately 60% to 70% of patients who present with colorectal carcinoma undergo surgical resection of the primary site; of these recurrence occurs in 30% to 40%. ²⁷ Aggressive monitoring with rapid identification of recurrent disease and early intervention improves survival in the postsurgical population. ²⁸ A summary of the factors complicating the evaluation of postsurgical patients follows, with a discussion of current advances in this field.

Bowel wall identification

A recent study by Morrin et al, ²⁹ in a series of 200 patients, demonstrated that IV contrast material significantly improved the delineation of the bowel wall, improving the sensitivity of the technique in cases where adequate bowel cleansing could not be achieved. Increased sensitivity (75%, compared with 58% in the noncontrast group) was present in the intermediate-sized (6- to 9-mm) polyp group. The technique described by Morrin et al ²⁹ involves the use of 150 mL of nonionic contrast in the prone position; rebolus was not given during supine imaging, to decrease patient exposure to contrast dye. An example of enhanced wall conspicuity is presented in Figure 3. ²⁹

Differentiation of lesions from normal colonic contours can also be problematic. Three-dimensional reconstruction aids in this distinction, as the normal haustral pattern is clearly visualized on reconstructed images and aberrations in the mucosal contour are more evident (Figure 4). ³⁰ Abnormal contrast enhancement can further serve to differentiate a sessile carcinoma from haustral marking, as demonstrated in Figure 5. It is important to consider, however, that the increased attenuation evident in lesions may be lost relative to fecal debris, should tagging be performed prior to the study.

Submerged polyps

Pooling of fluid within the bowel is another problem that can obscure polyps. Prone and supine imaging has been demonstrated to improve sensitivity significantly, in part by the movement of fecal debris and also by repositioning intraluminal fluid. Often, despite repositioning, fluid is present in sufficient quantity to prevent visualization of the full circumference of the bowel. As this fluid is typically of intermediate density and mixed with fecal debris, differentiation between it and a polyp is difficult. Lesion enhancement can improve sensitivity in these instances, by increasing tissue contrast. In the aforementioned study by Morrin et al, ²⁹ 2 larger lesions (>10 mm) obscured by fluid were identified only after contrast material administration (Figure 6). ²⁹

Adherent stool

Differentiation between adherent stool and polypoid lesions is also difficult. Although prone and supine imaging has improved the ability of virtual colonoscopy to distinguish between these, adherent stool often remains on the nondependent bowel wall. Contrast enhancement can help differentiate the two. In a study by Morrin et al, ²⁹ lesion conspicuity, particularly in the evaluation of smaller polyps (6 to 9 mm), was significantly increased after contrast material was administered. ²⁹

Postsurgical changes

Scar tissue often distorts the most critical region for postoperative surveillance, the resection bed. Patients may develop strictures, either at the site of anastomosis or from recurrent or metachronous disease. ^20,31,32 Contrast-enhanced CT plays an important role in separating malignant from benign stricture at the site of anastomosis; recurrent disease is often manifested by an irregular enhancing mass. Figure 7 demonstrates abnormal enhancement in recurrent carcinoma at the resection site. ²³

Bowel distention

Obtaining appropriate luminal distention is another significant concern. Bowel spasm may limit distention of some segments. Glucagon has been used by some institutions in these situations, although recent studies suggest that its utility is limited, except in patients suffering severe spasm. ³³

Patients with a right hemicolectomy often have ileocecal valve resection with consequent rapid reflux of air into the small bowel, which complicates necessary colonic distention. As previously discussed, IV contrast material can increase conspicuity of lesions in suboptimally distended bowel, reducing the effect of bowel decompression. ³⁰

Extracolonic disease

Looking beyond the importance of this method in evaluating colon cancer, and metastatic disease involving nodes, liver, lung, and other organs, evaluation of extracolonic disease is a significant benefit of virtual colonoscopy. Many studies have demonstrated the ability to identify renal cell carcinomas, and other asymptomatic diseases, whose early identification could result in curative therapy. For example, there has been a marked increase in the percentage of renal cell carcinoma incidentally detected, from 13% detected in 1985 to 73% in 1993. ³⁴ In a study of 264 patients, 11% of patients had significant extracolonic findings, 6 of whom required surgical therapy. ³⁵ This has been an increasing trend, as CT utilization has increased, and will continue to be an added benefit of this method to the community.

Current methods

The protocol at our institution involves 2.5 mm collimation, pitch of 1.5, 120 mA (varying with patient size), 140 kVp, and a 512 * 512 matrix, with 1-mm reconstructions. Images are obtained in both prone and supine positions. These images are transferred to our workstation, which uses a proprietary software package to reconstruct the images in three dimensions (Figure 8). This software then requires the input of several seed points to calculate the endoluminal view, which is represented in Figure 9. This data is then used to reconstruct the colon in which a strip of the entire colon is illustrated for rapid analysis (Figure 10). Any point on these displays can be selected, and the corresponding section on all images (cross-sectional, endoluminal, colon strip) is highlighted (CT Colonography package, GE Medical Systems, Milwaukee, WI).

Future directions

A recent mathematical model
comparing the cost-effectiveness of conventional colonoscopy and CT colonography suggests that at the present time, the conventional technique is a more economical option. ³⁶ Certain recent advances must be considered, however, in evaluating this technique. Multislice technology has dramatically decreased scanning time, enabling prone and supine images to be obtained in a total scanner time of fewer than 10 minutes. Image reconstruction has had a significant impact on the speed with which images can be interpreted. Most studies can now be evaluated in as little as 15 minutes. ³⁷ The focus of research at our institution is reducing this interpretation time even further. The software package (developed by Allen Tannenbaum, PhD, and colleagues, Georgia Institute of Technology) that is used reconstructs the colon from axial images, flattens it with minimal loss of contours, and thus allows immediate visualization of the entire circumference of the large bowel. This program is now under development to highlight suspicious regions automatically (A. Tannenbaum, unpublished data). ³⁸ This permits evaluation of the entire colon at a glance, calling attention to areas of concern, and greatly reducing interpretation time. Figure 11 illustrates a raw cross-sectional image. A thresholding algorithm is then applied to select out the colon (Figure 12). The colon is then reconstructed (Figure 13) and flattened (Figure 14). Color coding of the image then displays the topography of the colon, calling attention to mucosal abnormalities.

Conclusion

Virtual colonoscopy holds great promise in improving the widespread acceptance of colon cancer screening. It combines the advantages of speed and sensitivity with increased patient comfort and can simultaneously evaluate intra- and extracolonic disease. These abilities, when added to the heightened sensitivity provided by the administration of IV contrast, suggest that virtual colonoscopy will be a useful tool in patients with known or suspected colon cancer. Furthermore, it provides an essential surveillance tool in the postoperative population, in which recurrence is commonplace and early recognition essential to improving outcomes.