CHICAGO – A large study on computed tomography (CT) colonography with computer-aided detection (CAD) – commonly called virtual colonoscopy (VC) – has shown that the procedure is “highly effective for finding colon polyps,” according to the National Institutes of Health (NIH; Bethesda, Maryland) and presented at the Radiological Society of North America (RSNA; Oak Brook, Illinois) annual meeting here.

According to the presenter, CAD can improve the performance of virtual colonoscopy by marking abnormalities on VC images for review by the radiologist.

“The performance of virtual colonoscopy continues to improve, and the exam will become a colorectal screening method more patients and doctors will find acceptable,” said the study’s senior investigator, Ronald Summers, MD, PhD, in a statement.

Summers is staff radiologist, chief of the clinical image processing service and chief of the virtual endoscopy and computer-aided diagnosis laboratory at the NIH clinical center.

CT colonography, a minimally invasive exam, does not require entering the colon with an endoscope and therefore is believed by some to offer benefits such as eliminating the risk of bleeding or tearing of the colon with the endoscope. Also, intravenous sedation to achieve conscious sedation, which is typical during traditional colonoscopy, is unnecessary.

According to the RSNA, the procedure is less costly than conventional colonoscopy and more convenient, as it takes 15 minutes or less.

“Virtual colonoscopy is a patient-friendly exam that many patients would prefer,” Summers said during a press conference at the McCormick Place convention center to present some of the findings.

That is important, he said, because colon cancer is the second-leading cause of cancer death in the U.S., and the American Cancer Society (ACS; Atlanta) recommends that people at average risk for colon cancer begin regular screening at age 50, but current compliance with this recommendation remains well below 50%, according to the RSNA. Many people do not like the discomfort and unpleasant bowel-cleansing preparation required for a traditional colonoscopy.

The NIH study was a multicenter trial conducted at Walter Reed Medical Center, the U.S. Naval Medical Center and the San Diego Naval Medical Center. The software used in the study was developed by the NIH.

The study enrolled 792 patients using virtual colonoscopy with CAD to detect precancerous, or adenomatous, colon polyps 8 mm and larger. Colon polyps are benign, but they can develop into colon cancer if not removed, which often is accomplished during traditional colonoscopy.

Gastroenterologists would argue that if polyps are found during VC, a traditional colonoscopy still is required to remove the polyps.

CT colonography produces 600 to 1,000 images per patient. Summers likened the interpretation process to the “needle in a haystack problem.”

With CAD technology, after the radiologist has interpreted the images, the computer reviews the images and marks abnormalities for the radiologist to review, RSNA said. CAD, it said, has the “potential to find polyps that a radiologist might miss.”

“We think that CAD will help improve the performance of virtual colonoscopy by reducing the perceptual error that can occur when radiologists have to read the large number of images in these studies,” Summers said.

With CAD, the researchers detected polyps in 89.3% of patients with polyps 10 mm or larger and 85.4% of patients with polyps 8 mm or larger. CAD’s false positive rates were 2.1 per patient for polyps 10 mm and larger and 6.7 per patient for polyps 8 mm and larger, which fell within the researchers’ acceptable limits.

Virtual colonoscopy is used to analyze the shape of the surface of the colon. Various shapes are analyzed, such as normal folds and other shapes that may be polyps. Polyps can also occur in the folds of the colon, Summers said. Those polyps 1 cm or larger are typically those that need to be removed, he added.

The preparation for the VC procedure involved a clear liquid diet, bowel cleansing, stool and fluid tagging, and the colon was infused with room air in order to distend it, Summers said during the press conference.

Just as with conventional colonoscopy, even with VC, Summers said, “There’s going to have to be a method of distending the colon because otherwise the two sides might be touching each other” and the sides of the colon can’t be seen. The VC was performed using a single breathhold.

An optical colonoscopy also was performed on the same day as the virtual colonoscopy.

Summers said that with the study, it was concluded that there was “no significant difference” between VC and conventional colonoscopy.

In the study, two cancers were identified with a rate of 0.7 false positives, he said. In fact, traditional colonoscopy missed one of the polyps until the VC procedure found it and then it was later seen with traditional procedure.

Using CAD, Summers said that he and the other researchers “believe that radiologists will find it easy to detect” polyps. “CAD works well for the most worrisome types and sizes of polyps,” he concluded, noting that VC is “likely to get even better.”

Responding to a question regarding prep for the VC, he agreed that bowel cleansing is the most bothersome aspect of traditional colonoscopy. However, he said that with traditional colonoscopy, patients must drink a gallon mixture in advance of the procedure, whereas VC patients only drink about three tablespoons of a cathartic.

“There are a small number of patients for whom [the] smaller prep would not be reasonable,” he said, such as those with certain pre-existing health conditions.

Summers acknowledged that CAD software is “not widely used” today.

Viatronix’ (Stony Brook, New York) V3DColon System was the first VC approved by the FDA for colon cancer screening in April 2004.

A study showing virtual colonoscopy as effective in cancer screening was published in the December 2003 New England Journal of Medicine, and another study reported in the April 2004 issue of the Journal of the American Medical Association found that VC did not find as many lesions as the “gold standard” of conventional colonoscopy.

In the JAMA study, published just prior to Viatronix’ FDA approval, a gastroenterologist at the Medical University of South Carolina (Columbia) reported that a total of 827 lesions were detected in 308 of 600 participants who underwent both procedures, with 104 patients having lesions of at least 6 mm.

But whereas standard colonoscopy detected 99% of lesions of 6 mm or less, VC detected only 39% at that size. And with lesions sized at least 10 mm, standard colon-oscopy detected 100% of lesions, compared to 55% using the virtual procedure.