| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Editorials |
1 From the Department of Radiology, Boston University School of Medicine, 88 E Newton St, Boston, MA 02118. Received September 8, 2004; accepted September 9. Address correspondence to the author (e-mail: joseph.ferrucci@bmc.org).
Recently, the New England Journal of Medicine and the Journal of the American Medical Association published widely contrasting results of large multicenter clinical trials on the use of computed tomographic (CT) colonography, also known as virtual colonoscopy (1,2). Striking success was reported for the first and much larger study by Pickhardt et al (1), and CT colonography was widely heralded as a major clinical advance. Findings in the report by Cotton et al (2) published 5 months later, however, indicated that the performance of CT colonography was relatively poor. Cotton et al (2) concluded that CT colonography with their techniques was not ready for general acceptance. Ransohoff (3), the author of an accompanying editorial, agreed and cautioned that the poor results reflected what might now be encountered for CT colonographic studies performed in community practice. The study published in the Journal of the American Medical Association, however, was led by a well-known gastrointestinal endoscopist, and the results provoked sharp criticism from leading radiologist researchers (4–7) in CT colonography, who promptly challenged the methods and validity of the findings of the study. Yet, 10 months later, the negative publicity for CT colonography that the report of Cotton et al (2) generated in the lay media and the uncertainty it created in the medical community at large remain largely unaddressed. Although a new large multicenter trial of CT colonography is being launched by the American College of Radiology Imaging Network, the results are not likely to be available for 2 years.
The appeal of CT colonography lies in the fact that it is a totally noninvasive type of CT examination in which no drugs, sedatives, or contrast media are used; thus, it provides an attractive alternative for the many patients who refuse to or cannot undergo colonoscopy. Although CT colonography is not yet reimbursable for the screening of asymptomatic patients, it is rapidly being adopted into radiology practice nationwide, mainly as a result of patient demand. When properly performed, CT colonography accurately shows almost all important polyps. Because it is an imaging test only and has no capability for removal of polyps, its potential role in colorectal cancer screening raises important questions about future testing strategies. CT colonographic imaging techniques have progressed since these two studies were completed, and further refinements are certain. Therefore, it is important to examine further the differences between these two studies in terms of technique, outcomes, and implications for colon cancer screening strategies.
A Tale of Two Studies
CT colonography is a technically demanding imaging technique in which the elements of success are multifactorial, that is, the devil is in the details. In the first of the two studies led by Pickhardt, a radiologist, CT colonography outperformed optical colonoscopy in a same-day back-to-back correlative comparison in 1233 asymptomatic subjects with average risk at three U.S. military hospitals (1). A well-conceived CT colonographic protocol was used by five experienced study radiologists, and their performance was compared with the performance of 17 colonoscopists. Data were analyzed on the basis of the total number of adenomas detected and the number of adenomas detected per patient, which is the more relevant index for recommendation of referral for colonoscopy. Sensitivity for adenomatous polyps was 94% for CT colonography versus 92% for optical colonoscopy at the 8-mm-diameter threshold size for polyps and 96% for CT colonography versus 88% for optical colonoscopy at the 10-mm-diameter threshold size for polyps. The accuracy of CT colonography for adenomatous polyps on a per-patient basis was 92% at the 8-mm-diameter threshold size for polyps and 96% at the 10-mm-diameter threshold size for polyps. CT colonography depicted 54 (91.5%) of 59 advanced neoplasms, whereas optical colonoscopy depicted 52 (88.1%). The negative predictive value of CT colonography was 99% for adenomas of 8 mm in diameter or larger, and the diagnostic performance of CT colonography was uniform across the three participating centers. In addition, a total of 55 polyps were missed at optical colonoscopy, and these polyps included one that was malignant and 11 mm in diameter.
The authors used state-of-the-art four- and eight-channel multi–detector row spiral CT scanners and a double dose (90 mL) of an oral colon cleansing preparation (Phosphosoda; Fleet Pharmaceutical, Lynchburg, Va). They, however, also used then-novel CT colonographic technology. This included preprocedure administration of diluted oral barium sulfate contrast material as part of the colon cleansing preparation to tag residual fecal matter. At the time of image interpretation, advanced viewing software was used to electronically subtract the opacified stool from the scan data, allowing faster reading and easier detection of true polyps. In addition, image interpretation was performed by using commercially available endoluminal three-dimensional fly-through movie loop software for interpretation of the images, that is, virtual endoscopy (Viatronix 3D Colon; Viatronix, Stony Brook, NY). A final key technique used for comparison of results involved the use of segmental unblinding during optical colonoscopy. With this method, the results of screening CT colonography were sequentially disclosed to the colonoscopist during the performance of colonoscopy so that any colon segment that showed positive CT colonographic results for polyps larger than 5 mm in diameter that were not seen at optical colonoscopy could be reexamined by the colonoscopist. This created an enhanced reference standard and allowed the results of virtual and optical colonoscopy to be independently validated. An immediate result was the conversion of 55 false-positive results to true-positive results at CT colonography.
Poorer results were recorded in the second and smaller study by Cotton et al (2), although their protocol included use of similar two- and four-section multi–detector row CT technology, same-day back-to-back comparison of virtual and optical colonoscopy, and similar segmental unblinding of results. Coinvestigator radiologists read the CT colonographic images, but their results were much poorer than those with optical colonoscopy or those with CT colonography obtained by Pickhardt et al (1). In a total of 608 patients, per-polyp sensitivity with CT colonography for detection of polyps that were 10 mm in diameter was only 52%; at a lower threshold size set at 6-mm diameter, sensitivity was only 23%. The sensitivity for detection in a patient with at least one polyp of 6 mm or larger in diameter was only 39%; at CT colonography, two malignant lesions were missed. Because the media reported this as a "failure of CT colonography," not surprisingly, radiologists experienced in CT colonography attempted to respond by pointing out what they believed were shortcomings of the study of Cotton et al (2) in letters to the editor that were published in the Journal of the American Medical Association (4–7). The criticisms were severalfold.
First, because the study of Cotton et al (2) had actually begun 2 years before the study of Pickhardt et al (1), older technology was used. The features of this technology were thicker lower resolution CT section collimation (in several sites) and older image visualization software, including use of primary two-dimensional rather than primary three-dimensional endoluminal interpretation. To what extent the older technology affected the performance of CT colonography in the study of Cotton et al (2) is unclear, but it probably did not help. Second, there was no documentation of the adequacy of preprocedure colon cleansing or air insufflation, key elements of proper CT colonographic technique. Third, no fecal tagging was used. Fourth, underestimation of the steep learning curve for CT colonography was also evident. For example, prior to commencement of the study, the radiologists were required to have read images in only 10 proved CT colonographic training cases, whereas the consensus among radiologists experienced in CT colonography is that images in at least 25, preferably 50, training cases be examined to acquire interpretative competence (4). Moreover, patient recruitment was slow and uneven, with only one of nine participating centers achieving sufficient experience (184 patients) to have an acceptable performance for interpretation. In that center, the sensitivity for detection of polyps that were 10 mm in diameter or larger was 82%, which more closely approached the results of the trial of Pickhardt et al (1). Overall, in 70% of the study patient volume, patients were seen by radiologists in the eight other centers, where the average case volume was only 54 patients. For these radiologists, the combined sensitivity for patients with polyps that were 10 mm in diameter or larger was only 24%. (For radiologists in the study of Pickhardt et al [1], it was 96%.) In addition to the very low case volume, the radiologists who participated in the trial of Cotton et al (2) were provided little continuous learning, as there was no provision for case-by-case reader feedback.
A final concern was that Cotton et al (2) reported their results by using two descriptors, which created a potentially negative bias against CT colonography. First, the polyp threshold size used for summary result reporting was set at only a 6-mm diameter. Because the sensitivity of CT colonography is known to decline with very small polyp sizes, the apparent performance of CT colonography thus worsened. With the 6-mm-diameter threshold size, the potential clinical significance of polyps of intermediate size (7–9-mm diameter) was thus exaggerated. In contrast, Pickhardt et al (1) chose the 8-mm-diameter reporting threshold size as more clinically relevant for reporting polyps of intermediate size.
The second difference was that in the study of Cotton et al (2) reporting of results involved description of detected abnormalities as "lesions." Because various small benign nonneoplastic mucosal elevations, such as hyperplastic tags and lymphoid follicles, were considered together with true adenomas, the performance of CT colonography suffered compared with that of optical colonoscopy, which performed well at helping to find and allowing removal of numerous tiny innocent nonneoplastic lesions. Once again, it is of note that Pickhardt et al (1) chose to report results by using the more specific terms adenoma versus nonadenoma. Tellingly, further analysis of data in their series revealed that 40% of polyps that were 6 mm or larger in diameter were nonadenomas (8). Thus, for these lesions, referral for colonoscopic polypectomy would have yielded little health benefit in terms of cancer prevention.
What Have We Learned?
Several lessons, conclusions, and questions emerge from comparison of these reports.
First is the issue of nomenclature and standards for reporting of results. Since CT colonography is an imaging technique largely developed by diagnostic radiologists but with very specific applications to optical colonoscopy and colon cancer prevention, all definitions, inferences, and conclusions will have to be very precisely drawn. Ideally, future trials with CT colonography will be designed with clearer agreement on outcome criteria. A blueprint for result reporting in clinical research about CT colonography has been offered by radiologists such as Dachman and Zalis (9).
Second is the steep learning curve apparent for CT colonography. Reader training and documentation of competence are now required of reader participants in ongoing trials about CT colonography (7). Ultimately, a formal credentialing process may be necessary for CT colonography to achieve credibility so that it can be used widely as a screening test. A useful precedent is breast cancer screening with mammography, for which a national accreditation program was legislated by Congress (Mammography Quality Standards Act) to assure quality services.
Third, how crucial to success is the technique of primary three-dimensional fly-through interpretation? Pickhardt et al (1) accorded it considerable credit for their success. However, there are no specific direct comparative data available. In fact, CT colonographic interpretations from the standard two-dimensional CT displays as used by readers in the study of Cotton et al (2) also are used by most radiologists experienced in CT colonography and have yielded excellent results in single-center reports (10–13). The answer to this question may never become clear and may not be that important, as software features and user choices will likely evolve further.
Fourth, there is the challenge to the concept of the advanced adenoma (adenoma with a diameter of 10 mm or villous or dysplastic components at histologic analysis) as the appropriate target for screening (14). Should a smaller polyp target size be established to exploit the potential benefits of polyp detection when large numbers of patients who might prefer to undergo CT instead of endoscopy undergo screening? Indeed, as noted previously, Pickhardt et al (1) chose a specific threshold size of 8 mm in diameter to report results for polyps of intermediate size rather than lumping together data for polyps of 7–9 mm in diameter and potentially obscuring information. Ransohoff (3), editorializing about the study of Cotton et al (2), also suggested the possibility of a reduced target threshold size. Thus, policies that would prompt recommendation of referral for optical colonoscopy with the 8-mm-diameter polyp threshold size at CT colonography seem increasingly likely in the future. At this size threshold, Pickhardt et al (1) calculate that approximately 15% of asymptomatic patients who undergo colorectal cancer screening with CT colonography would be referred for colonoscopy and polypectomy, although at the 10-mm-diameter threshold size, it would be 7.5%.
Fifth, the terms polyp and lesion commonly used as descriptors by the colorectal cancer screening community may be acceptable for clinical decision making but misleading when used to assess utilization, outcomes, and costs of colorectal cancer screening. The many diminutive nonneoplastic polyps or lesions now so often found and removed at colonoscopy pose no patient threat but add substantial system costs. Even the threat posed to the patient by small adenomas has been challenged (15,16). Hence, some utilization patterns of colonoscopy in clinical practice recently have been described as excessive, inappropriate, and unnecessary (17). Currently, insurance carriers cannot easily track real cost benefit yields of colonoscopic polypectomy, as incoming provider claims carry only Current Procedural Terminology and International Classification of Disease, Ninth Revision, codes, with no data required about polyp size or histopathologic findings (17). Because strategies for colon cancer prevention rely primarily on detection and removal of medium and large adenomatous polyp precursors, these might be a more useful focus of both data analysis and resource utilization (18) and, possibly, even reimbursement.
Sixth is the value of the negative CT colonographic result. For the patient, a key aspect of the appeal of CT colonography is that it offers a negative predictive value of approximately 95% for lesions that are of the 10-mm-diameter threshold size (1,10–13). Thus, the patient who undergoes CT colonography can be confidently reassured in 10 minutes, during which he or she is awake and does not require an injection, that he or she does not harbor lurking colon cancer. A "clean bill of health." No need to worry for a few years. Back to work!
Current Outlook for CT Colonography
The momentum for CT colonography adoption is recovering. Specific temporary Current Procedural Terminology codes for both screening (0066T) and diagnostic (0067T) CT colonography became effective in July 2004, although only for data tracking at present. Pilot reimbursement for patient screening with CT colonography already has begun at three health maintenance organizations in Wisconsin. Most U.S. hospitals now possess the appropriate modern multi–detector row CT technology to perform CT colonography, and hospital administrators realize that there are few entry costs for adding this new service (19). Many community hospitals are feeling competitive pressures to offer CT colonography, and gastroenterologists are being encouraged to lend endorsement. A broader choice of commercial preparationless reduced-laxative products are being introduced to improve patient comfort during colon cleansing. Computer-aided detection artificial intelligence software is being developed to improve the speed and the reliability of the actual interpretation of these large and mostly normal CT data sets. Professional quality of performance of CT colonography is being addressed by the American College of Radiology, which is developing a practice standard, and the Working Group for Virtual Colonoscopy, a confederation of researchers in CT colonography that is drafting consensus statements about practice patterns, a reporting schema, and guidelines for training. Ultimately, some envision an accreditation program as well. Activist radiologists are also asserting that on the basis of the available published evidence, CT colonography should now replace the double-contrast barium enema examination as an approved colorectal cancer screening test (20).
For patients who are unwilling or unable to undergo optical colonoscopy, CT colonography is a "compliance enhancer." Hospital industry data suggest that as many as 60% of patients who undergo CT colonography have never had any previous colorectal screening (19). In many areas of the country, colonoscopy volume outstrips manpower resources, with several months of patient appointment waiting times. Prescreening with CT colonography to select the relatively few patients with large adenomas for referral to colonoscopic polypectomy should produce better use of colonoscopy resources, as the overall case mix for colonoscopy will be enriched by including more patients with actionable lesions. In this sense, all participants—patients, radiologists, and colonoscopists—are winners.
FOOTNOTES
Authors stated no financial relationship to disclose.
REFERENCES
This article has been cited by other articles:
|
|
 |
|
  S A TAYLOR, D BURLING, M RODDIE, L HONEYFIELD, J MCQUILLAN, P BASSETT, and S HALLIGAN Computer-aided detection for CT colonography: incremental benefit of observer training Br. J. Radiol., March 1, 2008; 81(963): 180 - 186. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. A Kuehle, J. Langhorst, S. C Ladd, T. Zoepf, M. Nuefer, F. Grabellus, J. Barkhausen, G. Gerken, and T. C Lauenstein Magnetic resonance colonography without bowel cleansing: a prospective cross sectional study in a screening population Gut, August 1, 2007; 56(8): 1079 - 1085. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. S. Lee, S. H. Park, E. K. Choi, S. Y. Kim, M.-J. Kim, K. H. Lee, and Y. H. Kim Colorectal Polyps on Portal Phase Contrast-Enhanced CT Colonography: Lesion Attenuation and Distinction from Tagged Feces Am. J. Roentgenol., July 1, 2007; 189(1): 35 - 40. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Slater, S. A. Taylor, D. Burling, L. Gartner, J. Scarth, and S. Halligan Colonic polyps: effect of attenuation of tagged fluid and viewing window on conspicuity and measurement--in vitro experiment with porcine colonic specimen. Radiology, July 1, 2006; 240(1): 101 - 109. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. A. Taylor, S. Halligan, A. Slater, V. Goh, D. N. Burling, M. E. Roddie, L. Honeyfield, J. McQuillan, H. Amin, and J. Dehmeshki Polyp Detection with CT Colonography: Primary 3D Endoluminal Analysis versus Primary 2D Transverse Analysis with Computer-assisted Reader Software Radiology, June 1, 2006; 239(3): 759 - 767. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
