Although colorectal cancer has major public health implications, many patients who are eligible for routine screening resist colonoscopy because the procedure is invasive and carries a risk of perforation. Virtual or computed tomographic colonoscopy (CTC) may offer an advantage over standard colonoscopy in these respects, but it is not known how well it performs in routine practice. Available studies report a wide range in sensitivity for detecting lesions 10 mm or larger in size. Generally, the ability to detect lesions greater than 6 mm is desirable. Cotton and associates conducted this multicenter study to compare the performance of CTC with that of standard colonoscopy in routine practice settings.
Participants 50 years and older with an indication for colonoscopy underwent cathartic preparation of the colon before CTC followed by regular colonoscopy. During the latter, results from the CTC were revealed segmentally to help the colonoscopist identify any discrepancies in the two examinations. This combination of screening test, segmental unblinding, and information from any further diagnostic testing was considered the criterion standard, with essentially 100 percent positive predictive value. The outcome measures were the sensitivity and specificity of CTC and colonoscopy with regard to lesions 6 mm or larger.
A total of 600 patients had both procedures, with 827 lesions detected. Of these, 79.1 percent were 1 to 5 mm in size, 14.4 percent were 6 to 9 mm, and 6.5 percent were at least 10 mm. Of the 173 lesions larger than 6 mm, 29 were advanced lesions found in 104 participants. CTC identified 41 participants with lesions 6 mm or larger, whereas colonscopy identified 103 participants. Of the 496 participants with smaller lesions, CTC identified 449 patients, while colonoscopy identified all of them. Other measurements consistently showed better detection with conventional colonoscopy. Positive predictive values for detecting lesions of at least 6 mm and at least 10 mm with CTC were 46.6 percent and 50.0 percent, respectively, compared with 100 percent for colonoscopy as part of the criterion standard. Negative predictive values for CTC were 87.7 percent compared with 99.8 percent with colonoscopy.
CTC had a low sensitivity for detecting lesions, whether they were 6 mm or 10 mm. No improvement in the accuracy of CTC, or “learning curve,” was evident as the study progressed. Patients did not reveal a preference for one procedure over the other, possibly because both require bowel preparation beforehand. At this time, there is no evidence that CTC is useful in routine practice. With further technologic advancements and improvement in methods of three-dimensional reconstructions, CTC may have broader application in the future.
editor’s note: This article is a reminder of the limitations of colon cancer screening.1 Clearly, virtual colonoscopy doesn’t perform as well as the conventional method, but when we are evaluating existing screening methods, accuracy, acceptability, and cost also need to be considered. Although colonoscopy is the criterion standard, its accuracy is unclear. Colonoscopy appears to detect only 90 percent of large lesions and 75 percent of lesions less than 1 cm.2 Even though a colonoscopic examination will miss a proportion of malignancies, in terms of detecting lesions, it is still the best available screening tool to date. However, because the costs of even conventional colonoscopic screening are prohibitively high, a more cost-effective screening option, notably annual fecal occult blood testing and flexible sigmoidoscopy every five years, is probably preferable.3 If advanced technology such as virtual colonoscopy brings improvement not only in acceptability and accuracy, it will increase demand, which, in turn, will require a rational plan for footing the bill.—c.w.