Performance of CT Colonography in Clinical Trials

The amount of data accumulated in trials of CT colonography (CTC) has greatly increased in the past decade. The information from these studies is shaping clinical practice and public health policy regarding screening for colorectal cancer (CRC). This article examines the performance of CTC in clinical trials for individuals at average risk and increased risk for CRC. It also addresses the efficacy of CTC after incomplete colonoscopy, when colon preparations are reduced or eliminated, and in academic versus nonacademic environments. The data suggest that CTC is effective especially for the detection of larger lesions and when more advanced imaging technology is used.

The performance of computed tomography colonography (CTC) in clinical trials has improved in the past decade and is now providing the basis for the recommendations for its use by various professional and payer organizations. However, these groups have come to different conclusions following review of the available data. For instance, the American Cancer Society, US Multisociety Task Force on Colorectal Cancer, and American College of Radiology in 2008 recommended CTC as an appropriate screening modality for average-risk individuals. In contrast, the Centers for Medicare and Medicaid (CMS) stated in a recent decision memo that the available evidence was insufficient to recommend CTC as screening modality for average-risk individuals, and instead concluded that it remain noncovered. The US Preventative Task Force recently determined that the evidence for screening with CTC was inadequate (grade I) to assess the benefits and harms of the test. These disparate recommendations, generally derived from the same available data, have resulted in mixed messages to patients, health care providers, and third-party payer organizations.

This article summarizes the performance of CTC in major trials for a variety of clinical indications. It focuses on major clinical trials of CTC for average-risk colon cancer screening, including issues of precision, sensitivity, and specificity. The CTC clinical trials for patients at increased risk for colorectal cancer are also discussed. Other topics include the data for CTC use after incomplete colonoscopy, reduced preparation or unprepped CTC, and academic versus nonacademic environments for interpretation of CTC studies.

General considerations regarding performance in clinical trials

There are several general considerations regarding assessment of performance of an evolving technology like CTC across different clinical trials. These must be taken into consideration to fully appreciate potential differences in results. First, the approach to performance of CTC has evolved with improved technology and understanding of how to use it appropriately. Second, there is variation in the years of experience of the investigators or observers in the trials, as well as in the way trials are designed to minimize interobserver variability. Third, the patient population included in the trial must be considered when reviewing the trial results. Some studies have focused on average-risk individuals, whereas others enhanced the trial population by selecting those with previous polyps and therefore a greater likelihood of having positive findings (but biasing the radiologists). It is therefore not surprising that the interpretation of available clinical trial results varies by specialty, society, or third-party payer.

Performance of CTC in patients at average risk for colon cancer

The first major study published in 2003 by Pickhardt and colleagues on this topic involved same-day CTC and optical colonoscopy, and used proprietary three-dimensional (3D) CTC reading techniques and endoluminal fly-though views. Three different medical centers contributed 1233 asymptomatic average-risk adult patients to the study. The mean age in the study was 57.8 years, and men represented 59% of the study population. Exclusion criteria for the study included a positive fecal occult blood test, anemia, rectal bleeding, history of polyps or cancer, or an optical colonoscopy within the previous 10 years. In a process termed segmental unblinding, colonoscopists were first blinded to the CTC results when encountering a new segment of bowel, but then the CTC findings were revealed after assessing each subsequent colonic segment. For large polyps (10 mm or >) the investigators reported a per-patient sensitivity of 93.8% and specificity of 96%. These percentages decreased when smaller polyps were considered. For 6-mm polyps the per-patient sensitivity was 88.7% and per-patient specificity was 79.6%. The 2 patients with malignant polyps in this average-risk cohort were detected by CTC. An 11-mm malignant polyp was missed on optical colonoscopy before the results of the colonic segment were unblinded.

Macari and colleagues published a case series in 2004 of 68 asymptomatic average-risk men who underwent CTC followed by same-day optical colonoscopy. In this study, polyps were grouped in size categories of 1 to 5 mm, 6 to 9 mm, and greater than 10 mm. A total of 98 polyps were described at the time of colonoscopy. CTC detected 21 of the 98 polyps. CTC did identify all 3 of the polyps that measured 10 mm or greater by conventional colonoscopy. The detection sensitivity decreased as polyp size diminished from 9/17 (52.9%) for 6- to 9-mm polyps and 9/78 (11.5%) for diminutive polyps (1–5 mm). The investigators concluded that CTC was a sensitive and specific test for polyps greater than 10 mm. This study may have been limited by using a CT with only 4-detector rows, two-dimensional (2D) rather than 3D views, and because the performance and interpretation of CTC and optical colonoscopy was carried out by a limited number of individuals.

A major study by Kim and colleagues in 2007 compared parallel screening cohorts to compare CTC and optical colonoscopy in a general screening population. They compared 3163 individuals in a traditional screening colonoscopy protocol with 3120 participants in a separate CTC program. Patients in this study were excluded for a history of polyps, a history of a bowel disorder, and hereditary nonpolyposis colorectal cancer syndrome. The primary outcomes of the study were the detection of advanced neoplasia (defined as advanced adenomas and carcinomas) and total number of polyps removed. In this study more advanced CT scanners were used (8- and 16-detector rows). There were slightly more men than women in both cohorts and there was no significant difference in age between the 2 groups. Ninety-eight percent of patients in both groups were asymptomatic individuals, and 92% of subjects had no family history of colorectal cancer. The investigators found that there were identical rates of advanced neoplasm (4%) and advanced adenoma (3%) detection in the CTC and optical colonoscopy groups. The investigators reported that 7.9% of patients were offered referral to optical colonoscopy because they had polyps greater than 6 mm. As expected, many more polyps were removed in the optical colonoscopy group (2434) than in the CTC patients subsequently referred for therapeutic polypectomy (561).

Cornett and colleagues in 2008 reported a cohort of 159 asymptomatic adults with polyps greater than 5 mm seen on CTC who underwent a planned therapeutic optical colonoscopy. In this study the endoscopists were aware of the findings of the prior CTC. Eight- and 16-detector row CT machines were used and, again, there were slightly more men than women in the study population (51%). The average age of participants was 59.3 years. Two hundred and thirty polyps were reported on CTC and a total of 359 polyps were identified at subsequent optical colonoscopy. CTC missed 6.2% of polyps greater than 9 mm and 18.2% of polyps 6 to 9 mm. Seventy-two percent of the polyps detected on therapeutic colonoscopy, but not on CTC, were less than 6 mm. The investigators reported a false-positive rate of 5% for CTC (normal colonoscopy despite CTC finding).

In 2008, Johnson and colleagues, in conjunction with the American College of Radiology Imaging Network (ACRIN), reported their findings of a multicenter trial of CTC versus optical colonoscopy in 2600 average-risk adults, with a primary end point of the detection of large (>10 mm) colorectal adenomas and cancers. Patients enrolled in this trial were all aged 50 years or older, reported no major bowel symptoms or diseases, and were otherwise considered normal candidates for screening colonoscopy. All studies were performed on 16-detector row machines. The average age of the participants was 58 years and, unlike other major CTC clinical trials, most of the study subjects were women (52%). In the per-patient analysis, the investigators reported a sensitivity of 90% for polyps greater than 10 mm and a specificity of 86%. The sensitivity and specificity of CTC for smaller lesions was less, as noted in all previous studies. The positive predictive value of CTC, or the proportion of patients with positive test results who were correctly diagnosed by traditional colonoscopy, for polyps greater than 10 mm was 23%. The negative predictive value for lesions of this size was 99%.

In one of the more recent trials performed in Germany, Graser and colleagues reported the findings in 311 asymptomatic average-risk adults who underwent 5 separate but parallel screening modalities for colorectal cancer. These included CTC, colonoscopy, sigmoidoscopy, and fecal occult blood/immunochemical testing. All patients had their colonoscopy performed after CTC in a segmental, unblinded fashion. The CT scanners used in the study had 64-detector row capability, which was higher than any of the previous major trials. The enrolled patients were more than 50 years of age and did not report any gastrointestinal (GI) warning signs (blood in stool, change in bowel habits, abdominal pain), and were excluded from the study if they had a history of inflammatory bowel disease or family history of colon cancer. Of the 307 who completed CTC and optical colonoscopy, a total of 221 adenomas were reported. Approximately two-thirds of the overall adenomas detected were 5 mm or greater. Similar to previous clinical trials, the per-polyp and per-patient sensitivity and specificity increased in concert with the increasing size of the lesions. For adenomas greater than 9 mm the per-polyp sensitivity was 93.9% and per-patient sensitivity was 92%. Of the 46 advanced lesions detected, almost 30% were less than or equal to 9 mm.

These clinical trials are summarized in Table 1 . There are many important similarities between these trials. The studies relied on traditional colonoscopy as their reference standard. All studies used bowel preparation and stool tagging to aid in the diagnosis of polyps on CTC. All but the 2 earliest studies used higher-resolution CT scanners. All studied provided a real-world group of patients who could be considered as average-risk colon cancer screening candidates. Most of the studies used several highly trained gastroenterologists and radiologists to carry out the performance and interpretation of CTC and optical colonoscopies. CTC was considered to be a simple test to carry out on widely available machines without the use of sedation.

Table 1

Only gold members can continue reading. Log In or Register to continue