1. Home>
2. Specialties>
3. Gastroenterology>
4. Summary and Comment

National CT Colonography Trial Results

CTC’s performance is good in terms of sensitivity but borderline with regard to specificity.

The U.S. Multisociety Task Force (MSTF) on Colorectal Cancer, the American Cancer Society, and the American College of Radiology recently added computed tomography colonography (CTC) to their joint colorectal cancer screening guideline as an option for screening average-risk persons at 5-year intervals (JW Gastroenterol Apr 4 2008). Prepublication of results from the National CT Colonography Trial, also called the American College of Radiology Imaging Network Trial 6664, was instrumental in the MSTF’s decision.

This trial involved 2512 asymptomatic patients who were scheduled to undergo routine colonoscopy at 15 U.S. centers. All patients underwent bowel preparation and stool and fluid tagging. CTC was performed, followed by colonoscopy later the same day; the colonoscopy data were used as a reference standard. Radiologists who interpreted the data were required to have performed at least 50 CTC examinations or to have attended a specialized CTC training session. After completing a qualifying exam, the 15 top-scoring radiologists were selected. Interpretations were made using two-dimensional primary imaging with three-dimensional problem solving in 1280 cases and using primary three-dimensional flythroughs in 1251 cases. Patients who had polyps 10 mm detected by CTC but not by colonoscopy were advised to undergo an additional colonoscopic examination within 90 days. CTC and colonoscopy results were considered matched if the CTC-measured tumor size was within 50% of the colonoscopically measured size.

The mean sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of CTC for detection of lesions 10 mm were 0.90, 0.86, 0.23, 0.99, and 0.89, respectively. Per-polyp sensitivity of CTC for polyps 5 mm, 6 mm, 7 mm, 8 mm, 9 mm, and 10 mm was 0.59, 0.70, 0.75, 0.80, 0.82, and 0.84, respectively. Sensitivity for detection of adenomas or cancers 5 mm, 6 mm, 7 mm, 8 mm, and 9 mm was 0.65, 0.78, 0.84, 0.87, and 0.90, respectively; specificity ranged from 0.86 to 0.89. The median size of neoplasms detected and missed by CTC was 10 mm and 6 mm, respectively. No difference in performance between the two- and three-dimensional software was seen. Adverse events included severe nausea and vomiting for less than 24 hours after CTC in one participant; hematochezia after snare polypectomy, requiring 2 days of hospitalization, in another; and hospitalization for Escherichia coli bacteremia 24 hours after both procedures in a third.

If all patients with lesions 5 mm on CTC were referred for colonoscopy, the referral rate would be 17%; a referral threshold 6 mm would lower the rate to 12%.

Comment: This study’s results are likely to dominate discussions of CTC performance for the foreseeable future. The principal reason the MSTF included CTC in its colorectal cancer screening guideline is that CTC’s performance was substantially better than that of double-contrast barium enema (DCBE), which is currently also included in the guidelines. (We expect only about 50% sensitivity for large adenomas with DCBE, and patients clearly prefer CTC over DCBE.)

However, these findings are less promising than the results of Perry Pickhardt’s group (JW Gastroenterol Dec 16 2003). Specifically, the numerical sensitivity for polyp detection is lower in the present study, and the specificity is substantially lower. In clinical practice, a specificity of 86% for polyps 1 cm is likely to cause several problems. First, the low positive predictive value expected in screening populations means that many patients in whom CTC shows large polyps will endure lengthy colonoscopies, during which frustrated colonoscopists will examine the designated sections of the colon multiple times to determine whether polyps are actually present. Second, the number of patients undergoing follow-up colonoscopies will increase substantially. Such low specificities would be unacceptable for noninvasive tests such as fecal occult blood testing (FOBT). Finally, the implications of false-positive results in imaging studies are quite different from those of false positives in noninvasive tests such as FOBT. A false-positive radiographic test often makes colonoscopists feel obligated to be certain no lesion exists. Thus, I predict that after a negative colonoscopy, many of these patients will undergo yet another CTC to determine whether the CTC result is a true or a false positive.

Can we expect the same level of performance in clinical practice? We should remember that the study radiologists underwent specific training, and then the lowest-scoring performers on prestudy testing were excluded from the trial. Thus, average performance in clinical practice might be worse than what was seen here.

Overall, the strengths of the study lie in its multicenter design and its large size. The CTC performance results are good from the perspective of sensitivity but borderline with regard to specificity, and CTC performance varied across the study sites. Indeed, the range of results among centers suggests that local monitoring of both CTC and colonoscopy performance might be the best way to guide the choice of tests.

— Douglas K. Rex, MD

Published in Journal Watch Gastroenterology September 17, 2008

Citation(s):

Johnson CD et al. Accuracy of CT colonography for detection of large adenomas and cancers. N Engl J Med 2008 Sep 18; 359:1207.

• Original article (Subscription may be required)
• Medline abstract (Free)