Gastroenterology
Volume 131, Issue 6 , Pages 2006-2009, December 2006

Colon Cancer Screening, Polyp Size, and CT Colonography: Making Sense of It All?

  • Don C. Rockey

      Affiliations

    • Corresponding Author InformationAddress requests for reprints to: Don C. Rockey, MD, University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75390-8887.fax: (214) 648-0274.

University of Texas, Southwestern Medical Center, Dallas, Texas

Article Outline

 

See “Yield of Advanced Adenoma and Cancer Based on Polyp Size Detected at Screening Flexible Sigmoidoscopy” by Schoen RE, Weissfeld JL, Pinsky PF, Riley T, on page 1683; and “Computed Tomographic Colonography: Assessment of Radiologist Performance With and Without Computer-Aided Detection” by Halligan S, Altman DG, Mallett S, Taylor SA, Burling D, Roddie M, Honeyfield L, McQuillan J, Amin H, Dehmeshki J, on page 1690.

A great deal is known about colorectal cancer screening, but at the same time, many questions remain. We know that colorectal cancer is one of the leading causes of cancer death in the United States and is therefore a major health care issue. We know that colon cancer arises from a canonical polyp to cancer sequence. We know that screening for colon cancer has the ability to reduce mortality from colon cancer.1, 2, 3 We know that not all eligible candidates undergo cancer screening and that many people have yet to be screened.4 We assume that screening the entire colon is important.5 We know that there are several potential methods to screen the colon, but we do not know which is the best. The strongest data come from trials using fecal occult blood testing.1, 2, 3 Although we lack unequivocal proof, it is believed that colonoscopic screening leading to removal of polyps reduces the incidence of colon cancer, and reduces colon cancer mortality.6, 7 However, colonoscopy is invasive, expensive, and really does not meet the criteria for an ideal screening test.

Over the last decade, computed tomographic (CT) colonography (also virtual colonoscopy) has been introduced as a novel method to examine the colon, and perhaps to screen it for colon cancer. This technology is attractive because it is simple to perform (for patients), probably very safe (although there are now reports of colon perforation, usually in elderly patients and those with underlying colon disease, and there is controversy about the risk of radiation exposure8, 9, 10), and in theory, should be accurate (see Rockey11 for further review). Although many issues have arisen, the major controversy in the field has surrounded the relative accuracy of CT colonography.12, 13, 14, 15, 16, 17 Several smaller studies involving populations at high risk for colorectal pathology suggested a high polyp sensitivity of CT colonography compared with colonoscopy (up to 100%; see Rockey11 for review). Larger, mostly single-center studies demonstrated widely variable detection rates. In these studies, the per-polyp sensitivities for lesions in different categories based on polyp size were the greatest for larger lesions and were in the following ranges: <5 mm (20%–60%), 6–9 mm (40%–80%), and ≥10 mm (60%–90%). The specificity of CT colonography varied as well, but was generally in the 90%–95% range. Most recently, 2 larger, multicenter studies have demonstrated that CT colonography was significantly less sensitive than colonoscopy,13, 14 whereas another demonstrated that CT colonography was more sensitive than colonoscopy for detection of lesions ≥10 mm.15

Two studies published in this issue raise further questions about the use of CT colonography for colorectal cancer screening.18, 19 One study, the data from which might be adapted to approximate a threshold screening approach, examined subjects from the Prostate, Lung, Colorectal, and Ovarian Cancer (PLCO) Screening Trial, a randomized, controlled, community-based study that used flexible sigmoidoscopy as an initial screen.18 The study cohort included 10,850 patients who had a polyp visualized on screening flexible sigmoidoscopy and who then underwent colonoscopy. Notably, at the time of sigmoidoscopy, lesions were sized visually but were usually neither biopsied nor snared. A lesion of any size identified at the time of sigmoidoscopy triggered colonoscopy. Of 1812 subjects with lesions ≥1 cm, 117 (6.5%) were subsequently found to have cancer in the distal colon, and 989 of 1812 patients had advanced adenomas (54.6%; defined as containing villous or tubulovillous histologic features, ≥1.0 cm in size, or had severe dysplasia). It should also be pointed out that reporting of histologic abnormalities was for subsequent findings at colonoscopy rather than for specific lesions. For those in whom the largest polyp size was 0.5–0.9 cm on screening sigmoidoscopy, 8 of 1426 women (0.6%) and 15 of 2183 men (0.7%) later had cancer identified by colonoscopy in the distal colon. In this group, 207 of 1426 women (14.5%) and 348 of 2183 men (15.9%) had advanced adenoma in the distal colon. Among the 3609 persons with the largest sized polyp 0.5–0.9 cm identified on sigmoidoscopy, 5.5% (198) subsequently had distal advanced adenomas that measured <1.0 cm but had villous histology or high-grade dysplasia, and 9.9% (357) had adenomas ≥1 cm found in the distal colon. A weakness of the study was that there was no matching of lesions, so that it is not possible to know if a lesion seen on sigmoidoscopy was also found on colonoscopy. However, in a subgroup of patients with only 1 lesion identified at each sigmoidoscopy and colonoscopy (therefore, presumably the same lesion), in those with initial lesions at sigmoidoscopy of 0.5 cm, 25.6% of these lesions measured ≥1.0 cm on colonoscopy and for those with initial lesions at sigmoidoscopy of 0.7 cm, 35.6% of these lesions measured ≥1.0 cm on colonoscopy.

In a second study published in this issue, investigators examined radiologists’ performance of CT colonography with and without computer-aided diagnosis (CAD).19 CT and clinical data came from 7 different centers and included patients examined for screening and investigation of symptoms. Because the authors were interested in understanding how those without experience in CT colonography would perform, 10 readers trained in CT but without experience in CT colonography were recruited to interpret CT colonography examinations in 107 patient data sets (including 60 sets that had 142 polyps), first without CAD and then with CAD (the investigators developed their own CAD program with the data set). Readers used MedicColon 1.2 (Berkeley Square, London, UK) software that displayed images in adjacent axial prone and supine views, along with multiplanar reformatting and a 3-dimensional rendered endoluminal view for problem solving. Per-patient and per-polyp detection were determined by comparing responses with known patient status. For polypoid lesions ≥1 cm, per-patient reader sensitivity among the 10 readers varied from 5/14 (36%, lowest) to 10/14 (71%, highest). For lesions 0.6–0.9 cm, reader sensitivity was 4/26 of 14 (11%, lowest) to 14/26 (54%, highest). For lesions ≤0.5 cm, reader sensitivity was 0/20 of 14 (0%, lowest) to 12/20 (60%, highest). With CAD, 41 (68%; 95% confidence interval [CI] 58%–77%) of the 60 patients with polyps were identified more frequently by readers and per-patient sensitivity increased in 70% of readers. However, overall performance was relatively poor even with CAD; on average, readers detected only 10 (51.0%) polyps ≥1.0 cm, and 24 (38.2%) ≥0.6 cm. Interestingly, reader time was reduced from 12.4 minutes without to 10.5 minutes with CAD. Notably, CAD itself generated many false-positive lesions (11.6 per patient).

These studies raise a number of important questions and issues. First, can polyp characteristics (in particular, those that could be reported by a screening imaging technique such as flexible sigmoidoscopy, CT colonography, wireless capsule endoscopy, “Aer-O-Scope,” or others20) be used to triage care? Evidence suggests that patients with adenomas can be stratified according to the risk for development of subsequent cancer. For example, those with multiple adenomas and those with large adenomas appear to be at the highest risk.7 In this regard, if ≥3 polyps are found at the time of CT colonography, these patients should be referred for colonoscopy to have lesions removed. Patients with large lesions should also be triaged for aggressive intervention.

Neither screening sigmoidoscopy as performed in the PLCO study nor CT colonography provided histology that could be used for further triage. On one hand, use of histology for triage is problematic because it is subjective, and there is great variability in classification schemes. On the other hand, there appears to be a relationship between the risk of colorectal cancer and increasing villous histology or carcinoma in situ compared with tubular histology.21 Unfortunately, we do not have enough natural history data to understand whether histology varies over time, and if so, by what degree. Finally, use of histology, obtained at the time of sigmoidoscopy, would affect cost, although the direction is not known (biopsy is costly, but savings could come from reductions in the numbers of referrals for colonoscopy). Thus, although it would be ideal for the clinician to be able to manage patients based on histology of polyps, we do not yet have the data that would allow this.

Thus, the question becomes, can polyp size determined by a screening imaging test (CT colonography, etc) be used to manage the patient with 1 or 2 “intermediate” (5–9 mm) sized lesions? Currently, there is agreement that lesions ≥10 mm identified at the time of CT colonography (or flexible sigmoidoscopy) should be pursued aggressively because of their higher risk of cancer. The smallest (<5 mm) lesions clearly have the least risk, and thus it may be safe to follow them for some period of time—perhaps 1–3 years—although the time intervals for follow-up are entirely unknown. Polyps in the 5–9 mm size range have a 0.5%–1.0% likelihood of harboring cancer, and an approximately 5% chance of high grade dysplasia.22 Additionally, the current PLCO study tells us that there is an even higher likelihood of finding lesions with advanced adenomatous features in the colons of patients with these “intermediate” sized polyps. What would happen to these lesions should they be simply followed? The answer, albeit unsatisfying, is that we simply do not understand the natural history of these lesions. It is likely that polyp biology varies from patient to patient, and probably even lesion to lesion. Thus, although many experts have made assumptions about polyp behavior and speculated about how they should be followed, the fact is that these recommendations are largely based on opinion.

Another critical point made by the PLCO study is that we are not very good at judging the size of polyps from 1 endoscopic examination to the next. Furthermore, matching of polyps seen at CT colonography and then subsequently at colonoscopy is also suboptimal (personal experience). The fact that we are so poor at polyp size assessment raises fundamental questions about the use of polyp size cutoffs to make clinical decisions. From a practical standpoint, if polyp size determinations are as variable as the data suggest, we must ask this: is it justifiable to use absolute size cutoffs to triage patients to different levels of care? It is this author’s belief that until we have better natural history data and can more accurately define polyp size, we should err on the side of the patient and investigate 5- to 9-mm lesions found on CT colonography or other imaging techniques.

The Halligan et al study19 emphasizes the point that the CT colonography field is a moving target. There is incredible variation in the way that CT colonography is and has been performed. This includes variability in preparation, use of air or CO2 during the examination, use of antispasmodics, and technical aspects in CT performance are possible (in collimation, slice thickness, reconstruction interval, table speed, acquisition time, and effective mAs), and in data management and software tools. Captured images have been visualized with software programs that depict data in any of a number of formats (multiplanar reformation, 2-dimensional, 3-dimensional rendering, virtual dissection, CAD). There has also been a great deal of variation in the types of patients studied (asymptomatic/symptomatic). This has led to wide discrepancy in reported outcomes (positive predicted value, negative predictive, and even sensitivity12), complicating interpretation of different reports. The study by Halligan et al19 is no exception, with use of multiple hardware platforms, different types of patients, and importantly, a distinct reading technique. Interestingly, sensitivities reported in this study were similar to those reported previously; the per-patient sensitivity for lesions ≥1.0 cm was 63% without and 76% with CAD. Indeed, the use of CAD in this study was noteworthy, but will it be consistent with other platforms so that we can generalize the results? All in all, it is clear that a major challenge in the CT colonography field is to develop consistency with regard to preparation, hardware, software, reading, and reporting.

Another important question is, can we really compare CT colonography and colonoscopy as colorectal screening modalities? And, additionally, how should they be used in clinical practice? It seems that a major problem in this area is that we have gotten into comparison shopping between CT colonography and colonoscopy. In fact, that may not be appropriate. CT colonography and colonoscopy both do image the colon, but in reality, they are different tests and have different strengths and weaknesses; we can appreciate these with the data in hand. Colonoscopy is mature, is highly consistent in terms of technique, is more accurate than CT colonography, can provide histology, and can remove precancerous lesions. CT colonography is immature, highly variable in terms of technique, is less invasive, is quicker to perform, but is less accurate and does not have the ability to provide histology or to treat disease. CT colonography also has the capability of detecting extracolonic lesions, which may or may not be beneficial. Thus, the discussion should probably turn not to one of “either/or,” but rather to how can we integrate these tests.

Patient preference for colon imaging procedures is a critical aspect of patient management in this field, particularly if patients require follow-up examinations. Several studies have examined patient experiences with CT colonography, and have compared the experience to that for other colon imaging tests (typically colonoscopy, but also air contrast barium enema).23 Some studies have demonstrated that patients have a preference for CT colonography, whereas others indicate that patients believe colonoscopy is preferable. All of these studies performed colonoscopy with sedation; it is possible that this specific intervention could have influenced patient preferences. “Minimally prepped” CT colonography offers a substantial advance in this area; this approach appears to be very well tolerated.24

Although technology such as CAD is attractive because it may increase reader sensitivity, it is also possible that it could increase the rate of false positives. In the study by Halligan et al,19 CAD itself had a false-positive rate of 11.6 per patient. It was therefore particularly surprising to see that the use of CAD reduced reader time; false-positive lesions must be reconciled carefully. It is further notable that not only do false positives lead to unnecessary colonoscopies, but they also incur an increased time commitment on the part of the endoscopist during colonoscopy.25

One of the more interesting aspects of the study by Halligan et al19 is that they recruited readers with minimal to no previous experience with CT colonography. They found that the reader sensitivity for CT colonography appeared to be with that reported previously, and overall, the per-patient sensitivity for lesions ≥1.0 cm was 63% without and 76% with CAD. Some would argue that these sensitivities are too low and that better training would have improved detection rates. Others would argue that these sensitivities are in line with published studies. In this case, the question becomes whether individuals with minimal previous experience can be trained to adequately read CT colonography.

If previous reading experience does not appear to be limiting in CT colonography, can nonradiologist readers, such as gastroenterologists, read CT colonography? From the work by Halligan et al,19 it appears that the learning curve is steep, suggesting rapid adaptation to the technology. However, although experience may be crucial, in 1 large trial, it was reported that those with less case experience actually appeared to detect more lesions than those with greater experience.13 This may have been due to the amount of time and/or degree of thoroughness taken during reading. Thus, some degree of experience will clearly be needed for nonradiologist readers to interpret CT colonography, but as for many procedures, experience is not all that is necessary; for CT colonography, meticulousness appears to be critical. Without question, we need further research to understand issues related to learning and training for many aspects of CT colonography.

Currently, some of those supporting CT colonography as a colorectal cancer screening tool have proposed that patients with polyps <10 mm in diameter be offered the option of ongoing CT colonographic screening.26 Others have suggested that this approach puts patients at risk and is not appropriate.22 The fact is that we currently just do not have the data to know what is right. What is clear is that not enough patients are being screened for colon cancer. It is also clear that health care costs are spiraling out of control—and it is obvious that novel technological advances, many of which are implemented without appropriate evidence supporting their use, are helping to fuel rocketing costs.

In the interests of our patients and our society, we should focus our efforts on attracting more patients into colon cancer screening programs. This means development of an integrated and organized approach to screening that is effective, yet cognizant of cost. Use of colon cancer screening tests, including, but not limited to, colonoscopy and CT colonography should be carefully optimized, not only in terms of their actual performance, but also in their implementation. It would seem that the time is right for the key stake holders (patients, physicians, payors, equipment manufacturers) to join together to meld new and old techniques in an effort to reduce the burden of colorectal cancer.

Back to Article Outline

References 

  1. Mandel JS, Bond JH, Church TR, Snover DC, Bradley GM, Schuman LM, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood (Minnesota Colon Cancer Control Study [published erratum appears in N Engl J Med 1993 26;329:672 (see comments)]). N Engl J Med. 1993;328:1365–1371
  2. Hardcastle JD, Chamberlain J, Robinson MHE, Moss SM, Amar SS, Balfour TW, et al. Randomised controlled trial of faecal-occult-blood screening for colorectal cancer. Lancet. 1996;348:1472–1477
  3. Kronborg O, Fenger C, Olsen J, Jorgensen OD, Sondergaard O. Randomised study of screening for colorectal cancer with faecal-occult-blood test. Lancet. 1996;348:1467–1471
  4. Seeff LC, Nadel MR, Klabunde CN, Thompson T, Shapiro JA, Vernon SW, et al. Patterns and predictors of colorectal cancer test use in the adult U.S. population. Cancer. 2004;100:2093–2103
  5. Schoenfeld P, Cash B, Flood A, Dobhan R, Eastone J, Coyle W, et al. Colonoscopic screening of average-risk women for colorectal neoplasia. N Engl J Med. 2005;352:2061–2068
  6. Selby JV, Friedman GD, Quesenberry CP, Weiss NS. A case-control study of screening sigmoidoscopy and mortality from colorectal cancer [see comments]. N Engl J Med. 1992;326:653–657
  7. Winawer SJ, Zauber AG, Fletcher RH, Stillman JS, O’Brien MJ, Levin B, et al. Guidelines for colonoscopy surveillance after polypectomy: a consensus update by the US Multi-Society Task Force on Colorectal Cancer and the American Cancer Society. Gastroenterology. 2006;130:1872–1885
  8. Sosna J, Blachar A, Amitai M, Barmeir E, Peled N, Goldberg SN, et al. Colonic perforation at CT colonography: assessment of risk in a multicenter large cohort. Radiology. 2006;239:457–463
  9. Burling D, Halligan S, Slater A, Noakes MJ, Taylor SA. Potentially serious adverse events at CT colonography in symptomatic patients: national survey of the United Kingdom. Radiology. 2006;239:464–471
  10. Brenner DJ, Georgsson MA. Mass screening with CT colonography: should the radiation exposure be of concern?. Gastroenterology. 2005;129:328–337
  11. Rockey DC. Colon imaging: computed tomographic colonography. Clin Gastroenterol Hepatol. 2005;3:S37–S41
  12. Mulhall BP, Veerappan GR, Jackson JL. Meta-analysis: computed tomographic colonography. Ann Intern Med. 2005;142:635–650
  13. Rockey DC, Paulson E, Niedzwiecki D, Davis W, Bosworth HB, Sanders L, et al. Analysis of air contrast barium enema, computed tomographic colonography, and colonoscopy: prospective comparison. Lancet. 2005;365:305–311
  14. Cotton PB, Durkalski VL, Pineau BC, Palesch YY, Mauldin PD, Hoffman B, et al. Computed tomographic colonography (virtual colonoscopy): a multicenter comparison with standard colonoscopy for detection of colorectal neoplasia. JAMA. 2004;291:1713–1719
  15. Pickhardt PJ, Choi JR, Hwang I, Butler JA, Puckett ML, Hildebrandt HA, et al. Computed tomographic virtual colonoscopy to screen for colorectal neoplasia in asymptomatic adults. N Engl J Med. 2003;349:2191–2200
  16. Johnson CD, Harmsen WS, Wilson LA, Maccarty RL, Welch TJ, Ilstrup DM, et al. Prospective blinded evaluation of computed tomographic colonography for screen detection of colorectal polyps. Gastroenterology. 2003;125:311–319
  17. Pineau BC, Paskett ED, Chen GJ, Espeland MA, Phillips K, Han JP, et al. Virtual colonoscopy using oral contrast compared with colonoscopy for the detection of patients with colorectal polyps. Gastroenterology. 2003;125:304–310
  18. Schoen RE, Weissfeld JL, Pinsky PF, Riley T. Adenoma and cancer based on polyp size detected at screening flexible sigmoidoscopy. Gastroenterology. 2006;131:1683–1689
  19. Halligan S, Altman DG, Mallett S, Taylor SA, Burling D, Roddie M, et al. Computed tomographic colonography: assessment of radiologist performance with and without computer-aided detection. Gastroenterology. 2006;131:1690–1699
  20. Bar-Meir S, Wallace MB. Diagnostic colonoscopy: the end is coming. Gastroenterology. 2006;131:992–994
  21. Loeve F, van Ballegooijen M, Boer R, Kuipers EJ, Habbema JD. Colorectal cancer risk in adenoma patients: a nation-wide study. Int J Cancer. 2004;111:147–151
  22. Van Dam J, Cotton P, Johnson CD, McFarland BG, Pineau BC, Provenzale D, et al. AGA future trends report: CT colonography. Gastroenterology. 2004;127:970–984
  23. Bosworth HB, Rockey DC, Paulson EK, Niedzwiecki D, Davis W, Sanders LL, et al. Prospective comparison of patient experience with colon imaging tests. Am J Med. 2006;119:791–799
  24. Iannaccone R, Laghi A, Catalano C, Mangiapane F, Lamazza A, Schillaci A, et al. Computed tomographic colonography without cathartic preparation for the detection of colorectal polyps. Gastroenterology. 2004;127:1300–1311
  25. Hur C, Gazelle GS, Hsu EH, Halpern EF, Podolsky DK. The effect of prior colonic imaging on endoscopic productivity: potential impact of computed tomographic colonography. Clin Gastroenterol Hepatol. 2005;3:1124–1127
  26. Pickhardt PJ, Taylor AJ, Kim DH, Reichelderfer M, Gopal DV, Pfau PR. Screening for colorectal neoplasia with CT colonography: initial experience from the 1st year of coverage by third-party payers. Radiology. 2006;241:417–425

PII: S0016-5085(06)02341-9

doi:10.1053/j.gastro.2006.10.065

Gastroenterology
Volume 131, Issue 6 , Pages 2006-2009, December 2006

Article Tools