Accuracy of CT Colonography for Detection of Large Adenomas and Cancers
Johnson CD, Chen MH, Toledano AY, et al
N Engl J Med. 2008;359:1207-1217
Much has been written recently about the emerging role of computed tomographic colonography (CTC) and its potential to replace conventional, optical colonoscopy. The long-awaited publication of the American College of Radiology Imaging Network (ACRIN) trial has brought a number of questions to the fore from both physicians and patients. This prospective trial involved 2600 asymptomatic study participants, 50 years of age or older, at 15 study centers. The CTC images were acquired with the use of standard bowel preparation, stool and fluid tagging, mechanical insufflation, and multidetector-row (16 or more rows) CT scanners. Radiologists trained inCTC reported all lesions measuring = 5 mm in diameter. Optical colonoscopy and histologic review wereperformed according to established clinical protocols at each center and served as the reference standard. The primary endpoint was detection by CTC of histologically confirmedlarge adenomas and adenocarcinomas (= 10 mm in diameter)that had been detected by conventional colonoscopy; detection of smallercolorectal lesions (6-9 mm in diameter) was also evaluated. The results for the diagnostic performance of CTC in polyp detection were as follows:
|= 5 mm||= 6 mm||= 7 mm||= 8 mm||= 9 mm||= 1 cm|
|Positive predictive value|
|Negative predictive value|
This discussion will focus not only on the ACRIN trial data but on emerging data on CTC as well. A key question to ask is: What potential impact might this new technology have on standard colonoscopic colorectal cancer screening? I will approach this discussion by addressing several focused questions.
1. What is the sensitivity/specificity of CTC compared with optical imaging?
Clearly, the data have been variable. Computed tomography has emerged as a potential alternative test for pancolonic imaging. The overall performance on the basis of data from the ACRIN trial was acceptable in regard to sensitivity, but specificity and positive predictive value were very disappointing. The specificity for detection of polyps > 1 cm in size was 86% and the positive predictive value was 25%. It is recognized that these data, although they are intended to represent a broad-based profile of academic and private practice, excluded 25% of the radiologists up front who did not achieve an appropriate score on a pretest qualification. Thus, the reliable extrapolation of these findings to interpreting physicians as a whole is very questionable and likely not reflective of what might occur in standard practice where physicians are not tested or excluded.
Given the above findings indicating a specificity of 86%, CTC performed every 5 years is almost an endorsement of screening colonoscopy. After 10 years and 3 cycles, 42% of the population will be colonoscoped for false-positive large polyp findings, let alone the percentage who will undergo colonoscopy for true positive findings and false-positive findings for which CTC suggests other polyps < 1 cm in diameter. The result will be an enormous increase in associated costs. Positive predictive values of 25% in clinical practice will be very problematic; colonoscopies will be prolonged, and I predict that most colonoscopies that are negative in a patient with a positive CTC will lead to the performance of another CTC within a year. Substantial percentages of the population will undergo 2 CTCs to prove that they did not have a polyp at the initial "screening" exam.
2. What does the evidence say about test characteristics for CTC variance as it relates to the type of scanning machine and software used, bowel preparation, reader training, and other operational factors?
Data for variability in CTC interpretation have been quite alarming. Johnson and colleagues conducted a single-center study at the Mayo Clinic in Rochester, Minnesota. Of 703 patients, the reported sensitivity among 3 experienced examiners for detection of polyps > 9 mm in diameter was 32%, 35%, and 73%, respectively, with a mean sensitivity of 46%. Cotton and colleagues conducted a large study at 9 expert centers in the United States involving 615 patients, with a primary endpoint of finding patients with polyps = 6 mm in size. They reported an overall sensitivity for this endpoint of 39%, with a sensitivity of 55% for identifying patients with polyps = 1 cm in size.
3. What does the evidence say about patients' attitudes and acceptance of CTC vs conventional colonoscopy for colorectal cancer screening?
Comparative studies of CTC and conventional optical colonoscopy have suggested a preference for optical colonoscopy (see table below).
Comparative studies of CTC and conventional optical colonoscopy have suggested a preference for optical colonoscopy (see table below).
|Reported Patient Experience With Different Colon Imaging Procedures|
|Air Contrast Barium Enema||Optical Colonoscopy||Computed Tomographic Colonography||P Value|
|Worried about the procedure||3.06||3.22||3.45||.0001|
|Directions difficult to follow||4.22||4.22||4.25||.94|
|Willingness to have test again||2.08||1.78||1.90||.0001|
|Worn out from procedure||3.22||3.03||3.57||.0001|
|Inconvenience of procedure||3.08||3.30||3.33||.0001|
Questionnaires were designed to score patients' level of agreement: 1 = totally agree; 3 = neutral; 5 = totally disagree.
Additionally, recent data have suggested a racial preference for optical colonoscopy over CTC. This study reported that racial/ethnic minorities were significantly less likely than whites to prefer CTC over optical colonoscopy for colorectal cancer screening (whites, 65.7%; blacks, 45.1%; Hispanics, 35.8%; and other, 35.7%; P < .001). Additionally, racial/ethnic minorities were less satisfied with CTC (whites, 8.4 ± 1.7; blacks, 7.8 ± 1.7; Hispanics, 7.4 ± 1.8; and other, 7.5 ± 2.1; P =.001) and were significantly less willing to undergo CTC again in the future (whites, 95.5%; blacks, 80.3%; Hispanics, 84.9%; and other, 85.7%; P =.006). Compared with white patients, optical colonoscopy was better tolerated and was preferred over CTC for the evaluation of the colon among racial/ethnic minorities. These findings suggest that CTC is unlikely to overcome racial/ethnic disparities in CRC screening.
4. What do the data say about the potential "harms" or adverse effects associated with CTC?
Reporting of polyps -- "ignoring based on size." Radiologists are currently discussing the value of a 1-cm cutoff size for CTC, in which polyps detected that are < 1 cm in size would either be ignored or followed up in a few years by repeat CTC. However, there is no evidence to suggest the safety of this approach. Overall, the literature suggests that polyps in the 6- to 9-mm range have an approximately 1% prevalence of invasive cancer at the time of discovery/removal, and it is unlikely that patients or primary care physicians will accept the strategy of leaving these polyps in place for 10-year intervals. Ignoring polyps that are < 1 cm in diameter represents an approach that has not previously been used in clinical practice and has never been the approach to polyps detected during a barium enema exam.
Furthermore, although there is a firm consensus that larger (= 10 mm in diameter) colonic polyps should be removed, the importance of removing smaller polyps (< 10 mm in diameter) is more controversial. In a recent analysis, investigators addressed the proposed strategy of CTC radiologists to "offer" colonoscopy for polyps 6-9 mm in size and to not report polyps < 6 mm. Closer interval surveillance would be included in this recommendation as well. If CTC is used for colorectal cancer screening, the majority of polypoid lesions identified will be < 10 mm in diameter. Decision-analytic techniques were used to compare the outcomes between 2 management strategies for smaller (6-9 mm) polyps detected on CTC. Hypothetical average-risk patients who had undergone a CTC examination and were found to have a small (6-9 mm) polyp were simulated to either: (1) undergo immediate colonoscopy for polypectomy (COLO), or (2) wait 3 years for a repeat CTC examination (WAIT). A Markov model was constructed to analyze outcomes, including the number of deaths and cancers after a 3-year follow-up period or time horizon. Values for the model parameters were derived from the published literature and from Surveillance Epidemiology and End Results (SEER) data, and an extensive sensitivity analysis was performed. The COLO strategy resulted in 14 total deaths per 100,000 patients compared with 79 total deaths in the WAIT strategy, for a difference of 65 deaths. The COLO strategy resulted in 39 cancers per 100,000 patients vs 773 in the WAIT strategy, for a difference of 734 cancers. Sensitivity analysis found that model findings were robust and only sensitive at extreme parameter values. The study authors concluded that managing smaller polyps detected on a screening CTC exam with another CTC examination 3 years later will likely result in more deaths and cancers than immediate colonoscopy and polypectomy. Given the recent robust data from the Centers for Disease Control and Prevention which suggest a decrease in colorectal cancer and related death, a test that ignores or minimizes the importance of lesions < 1 cm in diameter poses a significant "threat" to the value of colorectal cancer screening.
What is the histology of diminutive polyps that may not be reported on CTC exam? A recent study emphasized the prevalence of clinically important histology in small adenomas. Data were reviewed retrospectively from 3291 colonoscopies performed on asymptomatic patients found to have an adenoma on screening with flexible sigmoidoscopy a few weeks before the colonoscopy or who had a family history of colorectal cancer. All polyps were excised endoscopically and sent for pathology testing. Specimens with advanced histology were confirmed by a second reading. Of the 3291 colonoscopies performed, 1235 yielded a total of 1933 small or diminutive adenomatous polyps. Advanced histology including carcinoma was found in 10.1% of small (5-10 mm) adenomas and in 1.7% of diminutive adenomas (= 4 mm). Carcinoma was found in 0.9% of small adenomas and in 0% of diminutive adenomas.
Rex and colleagues evaluated a 5-year experience with 10,780 polyps removed from 10,034 patients at the University of Indiana. High-risk adenoma findings were defined as an advanced adenoma (= 1 cm in size, high-grade dysplasia, or villous elements) or 3 or more adenomas of any size, as per postpolypectomy surveillance recommendations. They found that 5079 (51%) patients had a least 1 polyp, 2907 (29%) had a least 1 adenoma, and 1001 (10%) had high-risk adenomas, which were either 3 adenomas = 5 mm in size and no polyps = 10 mm in size (or both findings). Of these patients, 18% had either 3 or more adenomas = 9 mm in diameter or an advanced adenoma = 9 mm in size (or both findings).
In this study by Rex and colleagues, of 2174 patients at least 50 years of age for whom screening was the primary indication, 15% had high-risk adenoma findings. Of these, 33% had either 3 or more adenomas = 5 mm in size or an advanced adenoma = 5 mm in size and no polyps = 6 mm. An additional 23% had no polyp = 10 mm in diameter, 1 or 2 polyps 6-9 mm in size, and 3 or more adenomas = 9 mm or advanced adenoma = 9 mm in size.
Overall, if the ACR CTC guidelines for reporting of polyps were applied to this cohort, and assuming 100% sensitivity for CTC in detecting polyps = 6 mm in diameter (clearly not the case), then 29% of all patients and 33% of screening patients 50 years of age and older with high-risk adenoma findings would be interpreted as normal, and an additional 18% to 23% of these patient populations, respectively, would have polypectomy delayed at least 3 years.
Risks for colon perforation associated with CTC. As for complications, is there really "virtually" no risk associated with CTC? In fact, there are 2 extremely large studies involving CTC that cite a perforation rate comparable to that of optical colonoscopy.[14,15]
Given that the numbers of complications reported for colonoscopy encompass all related diagnostic and therapeutic procedures (eg, polypectomy), these complications need to be put into proper perspective, especially when weighed against the use of alternative interventions (surgery) vs no intervention.
In the first study, researchers reviewed data from 11,870 CTC studies performed at 11 centers in Israel during a 2-year period. These studies represented more than 95% of all CTC studies performed in Israel during this interval. Seven perforations occurred, yielding a rate of 1 in 1700 (0.06%). Five perforations occurred in the sigmoid colon and 2 in the rectum. Four patients required surgical treatment.
In the second study, researchers interviewed the lead gastrointestinal radiologists at 50 centers in the United Kingdom to determine the number of CTC exams ever performed along with the number of perforations that occurred. Of all 17,067 patients (symptomatic and asymptomatic) who underwent CTC, 9 had perforations (1 in 1900, or 0.05%).
Overall, the perforation rate among symptomatic patients in these studies ranged from 0.03% (1 in 3400 patients) to 0.06% (1 in 1700 patients). These rates are much higher than those seen with barium enema studies and, for many individual endoscopists, exceed the rates of diagnostic perforation that occur during conventional colonoscopy.
CTC-related radiation risks. Most of the quantitative information on radiation risks and radiation-induced cancers comes from studies of survivors of the atomic bombing in Japan in 1945. This cohort of patients has been studied over time. Of note, there was a significant increase in the overall risk for cancer in patients/survivors who received "low-dose" radiation, ranging from 5 to 150 mSv. The mean dose of radiation for this group was approximately 40 mSv, which approximates the relevant organ dose exposure from a typical CT study in which 2-3 scans are done in a patient. Concerns were further corroborated by a recent study involving over 400,000 radiation workers in the nuclear industry. This study also reported a significant association between radiation dose and mortality from cancer. The range of exposure was between 5 and 150 mSv. These risks were quantitatively consistent with those of the atomic bomb survivors.
Results of prospective studies for patients who undergo CT scanning will not be available for a long time, but it is possible to estimate the risks of exposure by estimating the organ radiation dose exposure and applying the organ-specific cancer incidence or mortality data derived from the studies of the atomic bomb survivors. Brenner and Hall estimated the lifetime risk of death from cancer that was attributable to a single "generic" CT scan of the head or abdomen. The risks varied depending on the age of the patient at time of exposure and the organ-specific dose exposure (see table below).
|Study Type||Relevant Organ||Relevant Organ Dose (mGy or mSV)|
|Abdominal CT scan (adult)||Stomach||10.00|
|Abdominal CT scan (neonate)||Stomach||20.00|
Although the doses are higher for the head CT scan, the risks are higher for abdominal scans because the digestive organs are more sensitive than the brain to the development of radiation-induced cancer. Extrapolating from these data provided, the risk for cancer-related death associated with 1 abdominal CT scan is 0.06% for a patient exposed at 25 years of age and 0.02% for a patient exposed at age 50. This risk is striking and apparent when looking at the lifetime radiation risk of two of the most common radiogenic cancers, lung and colon cancer. For exposure to as little as 10 mSv at 25 years of age, the risk for death from lung and colon cancer is .025% and 0.0125%, respectively. For a patient exposed at 50 years of age, the risk is 0.017% and 0.010% for lung and colon cancer, respectively.
If it is valid to say that clinicians are not well versed in radiation risks as they relate to CT screening, it would not be surprising if these risks were not explained clearly to their patients before obtaining consent for an examination. However, by comparison, the estimated risk for serious complications and death associated with iodinated intravenous CT contrast is approximately 1 in 400,000, which is lower than the lifetime attributable risk due to single 10-mSv dose of radiation. Despite this, considerable attention is paid to the risk associated with the use of contrast during the consent process. This difference may be explained on the basis of a clear causal relationship: Contrast is injected and the patient immediately develops symptoms. By comparison, radiation effects may not manifest until 5-20 years after the exam, and causal relations are not apparent on an individual basis.
5. Finally, what do the data say about the cost impact associated with CTC?
With the frequency of repeat CTC exams, there is considerable concern over the cost-effectiveness of this approach. In a report by Kim and colleagues, further testing was recommended in 7.7% of CTC studies for indeterminate and probably unimportant or incompletely characterized lesions. The cost impact associated with additional testing for incidental findings, the direct risks associated with these tests (eg, possible biopsies), the indirect costs (eg, patient-related stress, work, quality-of-life issues), the extended risks of further radiation exposure, and the long-term related cost consequences of radiation-related complications all raise serious concerns about the additive and attributable costs of CTC.
The cost-effectiveness of CTC has not been well studied, particularly in light of the recent ACRIN data that reveal a relatively poor specificity that will lead to colonoscopy in 14% for each 5-year cycle, meaning that after 10 years (0-, 5-, 10-year screenings) 42% of patients would have been referred for colonoscopy (for suspected polyps = 10 mm) due to false-positive results.
Additionally, patient and physician behavior will be questioned in regard to follow-up if the CTC exam is positive and the colonoscopy is negative. Will patients worry and doctors order another CTC exam? Moreover, the costs -- both direct and indirect -- of incidental findings (extracolonic) on CTC exam and the related costs attributed to the testing strategy must be calculated and summed into any cost-effectiveness evaluation. A Washington State Health Care Authority analysis found that the cost of extracolonic findings alone were $2-$34 per screening. Indeed, extracolonic findings are important to consider given that a meta-analysis of 17 CTC studies found that 40% of patients had extracolonic findings and 14% underwent additional investigation because of these findings. The pursuit of extracolonic findings represents not only additive costs but also a potential additive risk to the patient attributed to ancillary testing (eg, radiation, biopsy, preparation). Any cost-effectiveness analysis of CTC must therefore take into account these extracolonic findings. With all of these additive costs, the "cost-effectiveness" of CTC will likely be hard to justify.
Coverage decisions for CTC as a screening study. Aetna recently issued a national noncoverage decision for CTC as a screening tool with very narrowly tailored exceptions. In its review of the evidence, the insurer stated, "More clinical trials need to be conducted to assess the cost-effectiveness and efficacy of virtual colonoscopy in comparison with conventional colonoscopy and sigmoidoscopy. The current cost of virtual colonoscopy probably prohibits its use as a screening tool." Humana also issued their coverage policy for CTC as a screening tool; they will only cover CTC as an alternative to colonoscopy when the patient is unable to complete or undergo a conventional colonoscopy exam due to a known obstruction.
The bottom line: Should we use conventional (optical) colonoscopy or CTC for colorectal cancer screening? The evidence speaks for itself! Although the potential exists for a favorable impact of CTC on CRC screening, it is not yet ready for "prime time."
Medscape Gastroenterology © 2008 Medscape
Cite this: CTC Screening (Virtual Colonoscopy): Is It Virtually Ready to Replace Optical Colonoscopy? - Medscape - Sep 26, 2008.