Computed tomographic (CT) colonography is a noninvasive method to evaluate the colon and has received considerable attention in the last decade as a colon-imaging tool. The technique has also been proposed as a potential primary colon cancer–screening method in the United States. The accuracy of the technique for the detection of large lesions seems to be high, perhaps in the range of colonoscopy. Overall, the field is rapidly evolving. Available data suggest that CT colonography, although a viable colon cancer screening modality in the United States, is not ready for widespread implementation, largely because of the lack of standards for training and reading and the limited number of skilled readers.
Computed tomographic (CT) colonography was initially proposed as a method to evaluate the colon for polyps and has grown and expanded remarkably over the previous decade. The growth of the new technology and research has been remarkable. There are many specific issues relevant to the test and its performance, which include the following: (1) test accuracy, (2) patient experience, (3) extracolonic lesions, and (4) colon preparation. There are also many new issues related to training. Also, whether the test is ready for widespread implementation is of critical importance.
The major attraction of CT colonography is that it is largely noninvasive (although there are growing concerns about radiation exposure, discussed later) and relatively simple to perform. Perhaps most importantly, because many patients are currently not being offered a colon cancer screening test, CT colonography offers an alternative, and data suggest that some patients who would not otherwise agree to screening may agree to undergo CT colonography.
The issue of CT colonography accuracy (ie, sensitivity to detect polyps and cancers) has evolved. Early studies examining highly selected populations and small cohorts, typically at high risk for colorectal cancer, revealed relatively high sensitivities and provided the foundation for the development of further studies. Later studies revealed wide variability in accuracy. Many larger single-center studies demonstrated extremely high sensitivities, but several well-publicized multicenter trials reported variable results ( Table 1 ). Two more recent multicenter studies incorporated elements that suggest that if generalizable, the technology and methodology is capable of detecting a high proportion of lesions. Although there are caveats, the field has advanced to the point that it should be expected that the sensitivity for detection of large (>1 cm) polyps is in the neighborhood of 90% and that this is a level that approaches that of colonoscopy. The caveats include questions about generalizability. For example, 2 of the studies seem to be more generalizable than the others. Also, for example, readers in some of these studies were highly trained, some studies used specialized preparative approaches, and some studies used specialized software platforms.
|Per Patient||Per Polyp|
|6–9 mm||≥10 mm||6–9 mm||≥10 mm|
|Pickhardt et al, 2003 a||(n = 168) |
|(n = 48) |
|(n = 210) |
|(n = 51) |
|Cotton et al, 2004||(n = 76) |
|(n = 42) |
|(n = 119) |
|(n = 55) |
|Rockey et al, 2005||(n = 116) |
|(n = 63) |
|(n = 154) |
|(n = 76) |
|Johnson et al, 2008 a||(n = 210) |
|(n = 109) |
|(n = 270) |
|(n = 76) |
|Regge et al, 2009 a||NA||(n = 131)||(n = 173)||(n = 174)|