Virtual colonoscopy, a less invasive method of detecting colon cancer, may not be ready for widespread use, according to a new study that found substantial variations in its accuracy. After comparing virtual colonoscopy with standard colonoscopy in several hospitals, the researchers found that the virtual procedure missed many precancerous polyps and even some cancers.

"The main result of this study was surprising and disappointing," wrote Peter Cotton, MD, in the Journal of the American Medical Association (Vol. 291, No. 14: 1713-1719). Cotton is director of the Digestive Disease Center at the Medical University of South Carolina, in Charleston, and was principal investigator for the study.

Some previous studies have suggested that virtual colonoscopy is comparable to standard colonoscopy, finding more than 90% of lesions in the colon. But those studies were all conducted in single hospitals by radiologists who were highly experienced with the technique.

The new study was conducted in 9 different hospitals by radiologists who were required to have performed at least 10 similar procedures previously.

"An obvious question is whether the radiologists in our study were sufficiently experienced [with the procedure]," said Cotton. "Our results indicate that [virtual colonoscopy] is not ready for routine use at this time, as many others have concluded."

Cotton and his colleagues compared the procedures in 615 patients. First, a radiologist performed virtual colonoscopy, which uses a computed tomography (CT) scanner to create images of the colon. Then a gastroenterologist administered conventional colonoscopy, which involves inserting a long, lighted tube into the colon to examine it with a small video camera.

In a side-to-side comparison, standard colonoscopy detected 99% of polyps 6mm or larger (103 of 104), while virtual colonoscopy detected only 39% (41 of 104). Standard colonoscopy detected 100% of polyps 10mm or larger (42 of 42), while the virtual colonoscopy found only 55% (23 of 42). Virtual colonoscopy also missed 2 of 8 cancers detected by conventional colonoscopy.

The accuracy of virtual colonoscopy varied widely between hospitals. One hospital that had a lot of experience with virtual colonoscopy had an 82% detection rate using this technique, but at the other 8 hospitals, the combined detection rate was just 24%.

These findings suggest that doctors who perform this procedure must be highly trained and experienced if they are to do it successfully.

"The differences between what virtual colonoscopy can do and what it will do if applied in ordinary practice circumstances are so great that physicians must be cautious," wrote David Ransohoff, MD, of the University of North Carolina, Chapel Hill, in an accompanying editorial.

Because virtual colonoscopy is less invasive than standard colonoscopy, it has been touted as having the potential to increase the persistently low levels of colon cancer screening in the US.

Only about half of the people who should be screened for this deadly cancer actually do so. Yet screening can detect colon polyps before they become cancerous, or find cancer in its earliest stages, when it is easier to treat. Colon cancer is the second leading cause of cancer death in the US.

"I hope our study will stimulate more studies on improving technique, software, scanners and training so that virtual colonoscopy will eventually provide an accurate, noninvasive screening tool," said Cotton. "If virtual colonoscopies are improved so they are as effective as the conventional colonoscopy, many more people would get screened, and more precancerous lesions would be detected."

Currently, patients must endure the same bowel cleansing preparation for both procedures, and if virtual colonoscopy does find a suspicious polyp, conventional colonoscopy is still needed to remove it.

Virtual colonoscopy could gain an edge, Cotton said, if techniques can be developed to eliminate the bowel preparation and simplify the process of interpreting the computerized images of the colon.

The American Cancer Society currently recommends that people at average risk for colon cancer begin screening at age 50. People are encouraged to have one of several tests, including traditional colonoscopy, flexible sigmoidoscopy, double contrast barium enema, or testing for blood in the stool.