Expert Commentary — Virtual Colonoscopy: Utility as a Screening Test for Colorectal Cancer?

Editor's Notes

Virtual colonoscopy represents a noninvasive test for the examination of the colonic lumen that involves the generation of both 2-dimensional and 3-dimensional views of the colon and rectum using data derived from computed tomography. This imaging modality has been explored as an alternative to conventional endoscopic colonoscopy, in particular as an alternative screening tool for colorectal cancer. Although virtual colonoscopy does not require sedation and requires less time for completion, the study requires the same bowel-cleansing preparation as conventional colonoscopy as well as gas insufflation of the intestine, which may be associated with patient discomfort.

In the December 4, 2003 issue of The New England Journal of Medicine, Pickhardt and colleagues reported results of the largest prospective evaluation to date regarding the utility of virtual colonoscopy as a screening tool for colorectal cancer. To explore the potential implications for clinical practice, Medscape asked 2 leading authorities to provide focused commentary to contextualize these study findings and offer perspective for both the gastroenterology and oncology communities, respectively.

How effective is virtual colonoscopy in screening for colorectal neoplasia when compared with conventional colonoscopy? Michael L. Kochman, MD, and Bernard Levin, MD, explore the issue.

Perspective for the Gastroenterologist: Commentary by Michael L. Kochman, MD

Introduction

The recent article by Pickhardt and colleagues[1] published in The New England Journal of Medicine has received a fair amount of attention due to the attractiveness of a “noninvasive” modality for the detection of colorectal neoplasia, within a colorectal cancer screening protocol. In this study, virtual colonoscopy demonstrated a sensitivity of 93.8% for adenomatous lesions 10 mm or larger, when analyzed per patient, vs 87.5% for conventional endoscopic colonoscopy. Only recently have most clinicians accepted the benefits of colorectal cancer screening programs,[2,3]and despite this fact, compliance with screening recommendations is not at the level that we as clinicians should accept.[4] There are a number of reasons for this suboptimal compliance with colorectal cancer screening guidelines, discussion of which is beyond the intended scope of this commentary. Suffice it to say that recommendations from physicians to undergo screening and patient adherence to these guidelines are necessary premises for a successful program.

Given this setting, the report by Pickhardt and colleagues certainly warrants commentary. The study authors are the first to demonstrate in a peer-reviewed article a sensitivity and specificity for virtual colonoscopy that is similar to that of endoscopic colonoscopy when analyzed for polyps >/= 10 mm. The methodology may be criticized on several counts, but most of these issues do not affect the overall findings. The technique of virtual colonoscopy that the study authors used is strikingly analogous to the performance of a per-oral pneumocolon and barium enema. The patients took oral barium and oral water-soluble contrast after a bowel preparation regimen. At the time of the virtual colonoscopy, the patients had a rectal tube inserted and “patient-controlled” insufflation occurred, although the exact parameters for the insufflation and the volume insufflated were not described. Not surprisingly, questionnaires assessing the associated comfort perceived by the patients revealed that individuals recalled discomfort more frequently with virtual colonoscopy than with endoscopic colonoscopy.

Methodology and Findings

The investigators discounted the finding of “nonadenomas,” an issue that is difficult to justify clinically, as histologic examination of a biopsy specimen is considered the gold standard in the United States for assessing the neoplastic potential of a colonic polyp. As demonstrated even in the present study, some “diminutive” polyps may contain unfavorable advanced and concerning histology. It is difficult to calculate the sensitivity and specificity for virtual colonoscopy overall, but when all identified polyps >/= 6 mm are evaluated, the specificity falls to 84.5% when analyzed on a per-patient basis. It is interesting to note that when the analysis is restricted to sensitivity, for adenomatous polyps >/= 6 mm, the rate for virtual colonoscopy also falls to 85.7% as opposed to the 90.0% seen with endoscopic colonoscopy.

The bowel preparation used in the present study exceeds that recommended for endoscopic colonoscopy and is not in keeping with the labeled use of the phosphosoda product, which limits it to 90 mL per 24 hours and does not include any additional laxative products. The Gastrografin administered (for opacification of luminal fluid) also has an additional laxative effect. Six patients were censured due to “failure of the CT colonographic system”; a recalculation of sensitivity and specificity on an intention-to-treat basis is not possible with these data as presented. Overall, only 6 patients were deemed to have poor preparations precluding complete evaluation; this is an astonishingly low figure, as the published numbers range from a low of 12% up to 30% in some series.[5] No clear scale for evaluating the quality of the colonic preparation was provided by the study authors, although given the computer algorithms used it would appear that in the setting of less than adequate colon preparations, an advantage would accrue to virtual colonoscopy. The overall completion rate for endoscopic colonoscopy was 99.4%. The study authors noted that the diagnostic performance of virtual colonoscopy was the same in all centers; no statistical information was given nor was an analogous statement made for the performance of the conventional endoscopic examination. The patient population studied included individuals with family history of colorectal carcinoma, which is somewhat bothersome because the a priori screening recommendations would be more frequent than every 10 years. One needs to be cognizant that recognized hereditary nonpolyposis colorectal cancer syndrome accounts for approximately 5% of all colorectal carcinoma in the United States and may present with sessile lesions.

A few caveats are present. Virtual colonoscopy does not allow for polypectomy or histologic evaluation of any findings; conventional colonoscopy would thus be required to confirm the results. The study authors suggest that small polyps may be followed with repeat virtual colonoscopy, although published data indicate that multiple small adenomas do increase the risk for additional advanced neoplastic lesions.[6] Published data and guidelines indicate that identified small polyps should therefore be biopsied and removed for histologic examination at the time of endoscopy. No data exist at this point concerning repeat screening intervals for virtual colonoscopy; therefore, further investigation is required to determine the clinical significance of the findings of a single polyp or multiple polyps < 10 mm in size by virtual colonoscopy.

Clinical Application?

There do appear to be current niche utilities for virtual colonoscopy. The gastroenterology societies and the published colorectal cancer screening guidelines do not include virtual colonoscopy as a standard modality.[2,3] The American College of Radiology has not yet published a white paper on standards of practice for virtual colonoscopy, including minimum training and equipment standards. It appears that the use of virtual colonoscopy is best reserved for those patients who are asymptomatic and who comprise low-risk populations that have failed traditional endoscopic screening (or absolutely refuse it) and who do not want an air-contrast barium enema.[7,8] Similarly, there may be utility in the preoperative use of virtual colonoscopy in patients with stenotic tumors in preventing the need for a preoperative total colonoscopy to rule out a more proximal second neoplasm, although no peer-reviewed papers indicate that virtual colonoscopy should replace colonoscopy post-resection.[9]

Concluding Remarks

At this point in time, the gold standard for colorectal cancer screening is still endoscopic colonoscopy. Primary-care physicians may feel that they are in a quandary, but there is clear guidance in the published literature. Additional studies need to be performed before routine application of virtual colonoscopy as a screening modality. Data do not support the use of this modality in the setting of gastrointestinal symptoms nor in patients known to have a history of colonic neoplasia. The currently available guidelines do not endorse virtual colonoscopy, and the US Preventive Services Task Force on Colorectal Cancer considers its use to still be in the research setting.[3] A formal cost-benefit analysis will need to be performed using the operating parameters as defined by the study authors and subsequent additional trials. Logically, however, it appears that overall healthcare costs may be increased and not decreased; the expected finding of neoplastic lesions will trigger endoscopic colonoscopy in a number of patients.

Virtual colonoscopy appears to be a promising imaging technique, and a larger body of literature must be accumulated. The development of rational, evidence-based algorithms may be warranted if there is confirmation of the results reported by Pickhardt and colleagues, in order to determine the precise operating characteristics of virtual colonoscopy in different settings. In the interim, it appears from all published data that any modality (fecal occult blood testing, sigmoidoscopy, and colonoscopy) that results in screening will at least have some positive impact on colon cancer rate reduction. One can assume, therefore, that virtual colonoscopy falls into this same category, although its exact clinical role is not yet clearly determined.

Perspective for the Oncologist: Commentary by Bernard Levin MD

Despite the attempts of media, sports figures, and professional societies to draw greater attention to the importance of screening for colorectal adenomas and cancer, adherence to recommended screening guidelines remains suboptimal.[4] In the past few weeks, the imagination of the general public and even physicians has been captured by a publication on the use of 3-dimensional virtual colonoscopy,[1] a noninvasive technique that might help boost screening rates in average-risk individuals and one that could help physicians better gauge who might benefit most from conventional colonoscopy.

In their study, Pickhardt and colleagues compared conventional colonoscopy with virtual colonoscopy in 3 medical centers. The sensitivity of virtual colonoscopy was greater than 93% for adenomatous polyps of 10 mm or longer. The average time spent by patients undergoing virtual colonoscopy was 14 minutes — about half that of conventional colonoscopy. The average time required for interpretation of each study was less than 20 minutes. Of note, extracolonic abnormalities of high clinical importance, including aortic aneurysms and certain cancers, were found in 4.5% of cases; other extracolonic findings of moderate clinical importance included nephrolithiasis and gallstones.

These performance characteristics are very promising. Indeed, if one were to use 10 mm as a cut-off of polyp size for conventional colonoscopy, 1 of 13 patients would be referred, making virtual screening a rather efficient tool for selecting out patients most likely to benefit from colonoscopy.

However, many questions have yet to be addressed. Can a wider array of community physicians achieve similar results? How will the public react when presented with the choice of yet another screening technique that still requires bowel preparation? Would making screening easier by offering conventional colonoscopy on the same day at a screening center be attractive or even feasible? What is the optimal interval between normal scans? How should we manage patients who have polyps that are smaller than 10 mm, for which the sensitivity for detection with virtual colonoscopy is lower? Can we devise a system or should physicians decide how to proceed after consultation with their patients on a case-by-case basis?

As a screening tool, virtual colonoscopy is not perfect and we clearly need additional data before advocating its widespread use. Yet it seems likely that this technique can enhance screening of those at average risk (ie, men and women 50 years and older), a population that might be more inclined to undergo a less invasive exam. Ultimately, though, as with other “high-tech” tools, adequacy of reimbursement and a competitive price will serve to reduce unnecessary obstacles to its use by the general public.