Introduction	Choose Top of pageABSTRACTIntroduction <<Subjects and MethodsResultsDiscussionReferencesCITING ARTICLES

Since its development by Vining et al. [1] in 1994, CT colonography has emerged as a potential screening technique [2–5]. Several studies have shown that CT colonography depicts colorectal neoplasms larger than 1 cm in diameter with sensitivity greater than or equal to 85% [6–8]. Colonoscopy is currently considered the reference standard for the detection of colorectal neoplasia in symptomatic patients and in the screening of high-risk asymptomatic individuals [9]. Conventional colonoscopy has various potential limitations. First, it fails to show the entire colon in about 5% of patients [10]. Second, it does not allow evaluation of the liver and other organs outside the colon. Third, it has a blind area, as a colonoscope passes in only one direction. For example, the opposite side of a colonic fold cannot be evaluated exactly. Finally, it is invasive and uncomfortable. CT colonography has been proposed as an alternative procedure for the examination of colorectal cancer, because it compensates for the limitations of colonoscopy [10]. Our principal focus in patients with colorectal cancer is the occurrence of cases of synchronous malignant lesions (1.5–9.0%) and polyps. Precise preoperative evaluation of the entire colon and precise staging are mandatory in the surgical approach and adjuvant therapies. MDCT has various advantages over single-detector CT, including improved image staging and polyp and cancer detection [11, 12]. However, in CT colonography studies, the use of 16-MDCT remains to be reported. The purpose of our study was to evaluate the usefulness of CT colonography for the staging of colorectal cancer and the detection of polyps and cancers in patients with the disease.

Subjects and Methods	Choose Top of pageABSTRACTIntroductionSubjects and Methods <<ResultsDiscussionReferencesCITING ARTICLES

Between October 2002 and August 2003, 51 consecutive patients (32 men and 19 women; mean age, 63 years; age range, 38–77 years) were prospectively included in this study. Patients were at high risk for colorectal cancer and had a history of altered bowel habits, anemia of unknown cause, abdominal pain, positive fecal occult blood results, and hematochezia. Eligible patients were informed of the study design and signed an institutional review board–approved consent form on which the procedure and study were explained. All CT scans were obtained on a 16-MDCT scanner (Mx8000 IDT, Philips Medical Systems) after a complete or incomplete colonoscopy. Colonoscopy was performed by a board-certified gastroenterologist. All masses and polyps identified on colonoscopy were photographed, measured, and resected at biopsy. Twenty-four hours before examination, each patient received a standard bowel preparation. Patients were allowed only a clear liquid diet until after the preparation had been completed. To prepare for the colonoscopy and CT colonography, all patients ingested 4 L of a polyethylene glycol electrolyte solution (Colyte, Schwarz Pharma), which was initiated at 6 pm on the day before the examination. Once the oral lavage solution had been administrated, each patient ingested 10 mg of bisacodyl, to reduce residual fecal material and retained fluid, at 10 pm on the day before the examination. Twenty milligrams of hyscine N-bromuro (Buscopan, Boehringer Ingelheim) was subcutaneously administered before air insufflation to further reduce bowel peristalsis and colonic spasm.

The scanning procedures were performed during a breath-hold. First, unenhanced images were acquired with patients prone; then, contrast-enhanced images were acquired with patients supine. The colon was insufflated by gentle squeezing of the enema bag, using room air, until the patients stated they were full or until 2,000 mL had been administered. The adequacy of the air insufflation was evaluated with a CT scout view, with more air insufflated if required. After scanning in the prone position had been completed, patients were positioned supine. After the administration of several additional puffs of air, the rectal tube was removed to improve patient comfort and prevent a possible rectal lesion. CT was performed after IV injection of an iodinated contrast agent, iopromide (Ultravist 300, Schering); 140 mL of contrast agent was administered at 3 mL/sec. CT was performed in the portal venous phase (start delay of 60 sec). The scanning parameters were 0.75-mm collimation, a pitch of 1–1.3, 120 kVp, 120–160 mAs, a 512 × 512 matrix, a 0.5-sec gantry rotation, and a 1-mm reconstruction thickness with 0.7-mm reconstruction intervals. The average acquisition time was 15 sec.

Images were processed on a PC, using commercially available software (Rapidia, Infinitt). The processed images included sagittal and coronal 2D reformatted, ray-sum, and virtual colonoscopy images. The evaluation began with 2D transverse CT images, followed by a virtual colonoscopy 3D view. The virtual colonoscopy viewing was performed both antegrade and retrograde, with the patient both supine and prone, to avoid blind areas.

The findings for each patient were prospectively recorded. The presence, location, size, and morphologic features of colorectal cancers and polyps were assessed. For TNM staging, the depth of tumor invasion and the presence of lymph nodes and metastases were evaluated. Tumor and node staging was based on the international TNM classification. Each tumor was assigned a stage, with T1 indicating a tumor in the mucosal or submucosal layer; T2, a tumor extending into but not beyond the muscularis propria layer; T3, a tumor extending beyond the muscularis propria layer into perirectal tissue and fat; and T4, a tumor directly invading other organs or structures. In CT image analysis, T1 and T2 were regarded as the same stage, because they are difficult to distinguish from each other on imaging. T3 was defined as a tumor with dense spiculation in the perirectal fat or a nodular or lobulated outer margin. For the node stage, N0 indicated no regional lymph node metastasis; N1, metastasis in one to three pericolic and perirectal lymph nodes; and N2, metastasis in four or more pericolic and perirectal lymph nodes. The number of positive lymph nodes was defined as the number of clustered nodes, independent of their size, or any lymph node measuring at least 1 cm in the long axis. M0 was no evidence of metastasis, and M1 was distant metastasis. The liver and both lower lungs were evaluated for metastasis. The diagnostic accuracy of TNM staging was calculated, and the sensitivity of CT colonography for the detection of cancers and polyps was compared with that of colonoscopy. All CT images were independently interpreted and evaluated by two abdominal radiologists who had at least 5 years' experience with abdominal CT and, to minimize any selection bias, were unaware of the colonoscopic findings. Differences in assessment were resolved by consensus. With regard to polyp detection, eight cases had to be resolved by consensus. The CT colonography results were compared with the surgicopathologic and follow-up colonoscopic findings, which served as the reference standard.

Eight colonic segments were considered to evaluate differences in overall colonic distention between the supine and prone positions: rectum, sigmoid colon, descending colon, splenic flexure, transverse colon, hepatic flexure, ascending colon, and cecum. We did not use objective criteria for colonic distention that have been defined in the literature [13]. Instead, colonic distention was ranked using the following three-point system: 1, suboptimal distention in more than three segments; 2, suboptimal distention in one to two segments; and 3, optimal distention visibility throughout. Optimal distention was defined as colonic wall and haustral folds that were “pencil-thin” throughout the segment. Suboptimal distention was an inadequate distention, with a slightly thickened colonic wall and haustral folds or lumen collapse.

Wilcoxon's signed rank test was used to evaluate the technical results for colonic distention. For detection of colorectal cancers and polyps, the statistical difference between the CT and initial colonoscopies was determined with Fisher's exact test. For all tests, a p value of 0.05 or less indicated a statistically significant difference.

Discussion	Choose Top of pageABSTRACTIntroductionSubjects and MethodsResultsDiscussion <<ReferencesCITING ARTICLES

The main benefits of CT colonography in colorectal cancer are its ability to evaluate the entire colon, to complete an unsuccessful colonoscopic examination, and to permit TNM staging. In our study, 11 patients underwent CT colonography for a distal occlusive carcinoma. In one of these 11 patients, a synchronous lesion was detected on CT colonography only. Other advantages of CT colonography over conventional colonoscopy include a shorter procedure, less risk to the patient, no need for IV sedation, and greater precision in localizing lesions. However, the disadvantages of CT colonography include the need for bowel cleansing, the long viewing and interpretation times, and the radiation dose.

MDCT has several advantages over single-detector CT, including increased temporal and spatial resolutions, faster data acquisition, and a wider field of view and comparable coverage times, with much thinner section collimation [14]. The use of thinner section collimation makes near-isotropic voxels for CT colonography. The advantages of an isotropic image include improved rates of polyp detection and accuracy of TNM staging because of reduced volume averaging and improved z-axis resolution for multiplanar reformations and 3D viewing. Compatible with Dukes' classification, TNM staging adds greater precision to the identification of prognostic subgroups. Although comparisons between 16- and 4-MDCT for CT colonography remain to be studied, 16-MDCT theoretically may yield greatly improved polyp detection rates, decreased false-positive findings, and more accurate TNM staging than does 4-MDCT. In our study, correlation with pathologic TNM stage was good, and sensitivities for the detection of polyps with a diameter of less than 5 mm, of 6–9 mm, and of more than 10 mm were 84%, 94%, and 100%, respectively. These results were higher than in previously reported studies using 4-MDCT [3, 11, 14]. Two additional reasons were thought responsible for the greater sensitivity of CT colonography over that of conventional colonoscopy for the detection of polyps. First, CT colonography could be used to evaluate the proximal colon beyond an obstructive lesion, whereas the colonoscope was unable to pass the obstruction. Second, CT colonography has no blind areas, such as the opposite side of the colonic fold, which colonoscopy cannot depict. In our study, the colonoscopic results were false-negative for seven polyps because of an occlusive carcinoma, with two polyps located near the colorectal cancer. These conditions were thought to be the main factors causing the false-negative results for polyps in patients with colorectal cancer. Also, 16-MDCT can reduce respiratory artifacts by reducing scanning time.

For polyp detection, interpretation technique is important. Review of transverse images as the primary interpretation technique is difficult and takes a long time for detection of small polyps. Smaller polyps might be quickly and easily detected through evaluation of virtual colonoscopy images. Therefore, transverse images and complete virtual colonoscopy images in both the antegrade and retrograde directions, with the patient both supine and prone, were used to increase the detection rate of small polyps. Although diagnostic accuracy for polyp detection was not compared between 2D and 3D interpretations, several polyps that were initially missed on the transverse image were later detected on the endoluminal image. No instance of the reverse occurred.

In this study, thinner-section MDCT was found to reduce false-positive results. The major causes of false-positive findings are residual fecal material, retained fluid, suboptimal bowel distention, and respiratory artifacts [15]. However, the combination of virtual colonoscopic images, transverse views, and multiplanar images was helpful in distinguishing residual fecal material from a colonic polyp. Internal heterogeneity was better depicted by the transverse image, and external morphologic features were better depicted by the virtual colonoscopy view. Morrin et al. [16] reported that contrast-enhanced CT colonography increased diagnostic confidence about medium-sized polyps (diameter, 5–9 mm). Residual fecal material can also be digitally subtracted from the images by dietary fecal tagging [17]. Although polyethylene glycol is highly effective at cleansing the bowel, it often leaves retained fluid in the colon. Bisacodyl decreases the residual fecal material and retained fluid. Retained fluid and collapsed segments have been markedly reduced by the combination of supine and prone imaging. No statistically significant difference in mean overall bowel distention was found between the supine and prone positions.

Our study has some limitations. First, in the TNM classification, a sensitivity of 100% was reported for detection of cancer. However, this percentage was inaccurate and potentially misleading. Eleven patients had obstructive lesions, and these were easy to detect. In such cases, the sensitivity is in localization of the lesion, not its detection. Second, the imaging criteria for TNM staging and the criteria for determining lymphadenopathy are unclear. Differentiation of T1 from T2 is also difficult. The number of patients with positive lymph nodes and metastasis was insufficient to determine the accuracy of CT colonography and should be addressed to evaluate colon cancer. In our study, only one patient had metastases. Third, no objective criteria were used for bowel distention. Taylor et al. [13] graded colonic distention as partial collapse (grade 1), suboptimal distention (grade 2), or optimal distention (grade 3) according to the thickness of the colonic wall and haustral fold. Optimal distention was defined as a colonic wall that was pencil-thin throughout the segment, with haustral folds less than 2 mm thick throughout their length.

In conclusion, CT colonography using 16-MDCT correlated well with pathologic TNM stage and was more sensitive than colonoscopy for the detection of cancers and polyps. Our preliminary data suggest that for patients in whom colorectal cancer is clinically suspected, CT colonoscopy would be valuable for staging the tumor and detecting polyps or cancers in areas not evaluated by conventional colonoscopy. This possibility should be confirmed by a larger study.