How accurate is computed tomographic colonography as a screening test for colorectal cancer?

Computed tomographic (CT) colonography will detect approximately 90% of large polyps and adenocarcinomas. The benefits include no need for sedation and a lower risk of perforation; the harms include the need for subsequent colonoscopy in approximately 1 in 6 patients, the varying expertise of radiologists, and exposure to ionizing radiation. The lifetime risk of cancer from the radiation of a CT colonography program is estimated to be 0.14%, or approximately 1 in 600 (Gastroenterology 2005;129:328-37) and must be balanced against the reduction in the risk of colon cancer and the risks of optical colonoscopy. The decision to adopt this technology on a large scale should include a careful and objective assessment of these benefits and harms. (LOE = 1b)

Johnson CD, Chen MH, Toledano AY, et al. Accuracy of CT colonography for detection of large adenomas and cancers. N Engl J Med 2008;359(12):1207-1217.  [PMID:18799557]

Cohort (prospective)

Government

Outpatient (specialty)

These researchers recruited 2600 adults, 50 years and older, who were free of significant bowel symptoms. Each had a standard bowel preparation, then underwent CT colonography using a 16-row scanner, and had an optical colonoscopy. Radiologists and endoscopists were masked to the results of the other study; radiologists were additionally randomized to reviewing either 2-dimensional images only or a combination of 2-dimensional and 3-dimensional images. If a lesion 10 mm or larger was detected on CT colonography but not on the initial optical colonoscopy, a second "unblinded" optical colonoscopy was scheduled 90 days later. The average age of the study participants was 58 years, 48% were men, and 9% had a family history of polyps or colon cancer. The sensitivity of CT colonography increased with lesion size; on a per patient basis, it was only 65% for detection of lesions of at least 5 mm, 84% for detection of lesions of at least 7 mm, and 90% for detection of lesions of at least 10 mm in diameter. The corresponding sensitivies on a per polyp basis were 59%, 75%, and 84%, respectively. Specificity was 86% for lesions of at least 10 mm, resulting in a positive predictive value of only 23% and negative predictive value of 99%. Using a cutoff of 5 mm for referral, 17% of patients would have been referred for follow-up colonoscopy, and approximately 1 in 4 would have actually had a large polyp or cancer. Sensitivity also varied considerably between radiologists, from less than 70% to 100%, and was not associated with the number of examinations.