GASTROINTESTINAL

Five year colorectal cancer outcomes in a largenegative CT colonography screening cohort

David H. Kim & B. Dustin Pooler & Jennifer M. Weiss &

Perry J. Pickhardt

Received: 11 October 2011 /Revised: 17 November 2011 /Accepted: 5 December 2011 /Published online: 31 December 2011# European Society of Radiology 2011

AbstractObjectives To assess the 5-year incidence of clinically present-ing colorectal cancers following a negative CT colonography(CTC) screening examination, as few patient outcome dataregarding a negative CTC screening result exist.Methods Negative CTC screening patients (n01,050) in theUniversity of Wisconsin Health system over a 14-monthperiod were included. An electronic medical record (EMR)review was undertaken, encompassing provider, colonoscopy,imaging and histopathology reports. Incident colorectal cancersand other important GI tumours were recorded.Results Of the 1,050 cohort (mean [±SD] age 56.9±7.4 years),39 (3.7%) patients were excluded owing to lack of follow-upwithin our system beyond the initial screening CTC. Theremaining 1,011 patients were followed for an average of4.73±1.15 years. One incident colorectal adenocarcinomarepresented a crude cancer incidence of 0.2 cancers per1,000 patient years. EMR revealed 14 additional patientswith clinically important GI tumours including: advancedadenomas (n011), appendiceal goblet cell carcinoid (n01),appendiceal mucinous adenoma (n01) and metastatic ileoco-lonic carcinoid (n01). All positive patients including theincident carcinoma are alive at the time of review.Conclusions Clinically presenting colorectal adenocarcinomais rare in the 5 years following negative screening CTC,

suggesting that current strategies, including non-reporting ofdiminutive lesions, are appropriate.Key Points• CT colonography (CTC) screening is increasingly used toidentify potential colorectal cancer.

• Clinically presenting cancers are rare for 5 years followingnegative CTC screening.

• The practice of setting a 6 mm polyp size threshold seemssafe.

• An interval of 5 years for routine CTC screening isappropriate.

Keywords CTcolonography . Gastrointestinal . Screening .

Adenocarcinoma . Advanced adenoma

Introduction

Computed tomography colonography (CTC) is a less inva-sive alternative method to colonoscopy. Several multicentretrials [1–4] have validated performance in large polyps andmasses, leading to inclusion in the American Cancer Societycolorectal cancer (CRC) screening guidelines [5]. CTC-based screening has been shown to be an efficient methodfor advanced neoplasia detection where yields equivalent tocolonoscopy-based programmes are seen, despite substantiallyfewer polypectomies [6].

On the other hand, critics contend that widespreadscreening by CTC may lead to interval cancers that couldbe avoided if screened by colonoscopy. They argue that theCTC selective polypectomy strategies where diminutivepolyps are ignored and surveillance of isolated (one ortwo) 6- to 9-mm polyps is permissible result in manyhigh-risk polyps and cancers that are not removed [7].Furthermore, such proponents assert that CTC is unable to

D. H. Kim (*) :B. D. Pooler : P. J. PickhardtDepartment of Radiology,University of Wisconsin School of Medicine and Public Health,E3/311 Clinical Science Center, 600 Highland Ave.,Madison, WI 53792-3252, USAe-mail: dkim@uwhealth.org

J. M. WeissSection of Gastroenterology, Department of Internal Medicine,University of Wisconsin School of Medicine and Public Health,Madison, WI, USA

Eur Radiol (2012) 22:1488–1494DOI 10.1007/s00330-011-2365-2

detect many flat lesions, which may hold higher rates ofhigh grade dysplasia and invasive cancer [8]. Thus, ifsuch hypotheses were correct, a negative result at CTCmay lead to excessive numbers of cancers presentingsymptomatically between routine screening intervals.The purpose of this study was to determine the rate ofclinically presenting incident cancers over a 5-yearfollow-up in a large negative CT colonography screeningcohort.

Materials and methods

This Health Insurance Portability and Accountability Act(HIPAA)-compliant study was approved by the institutionalreview board at the University of Wisconsin School ofMedicine and Public Health. The need for informed consentwas waived.

Study group

Between April 2004 and May 2005, 1,209 consecutiveindividuals were enrolled in the University of Wisconsin(UW) CT colonography screening programme. Participationrequires a physician/medical care provider referral; theprogramme is not yet an open access model. Individualsare generally at average risk of colorectal cancer and referredby their primary care provider for the purpose of colorectalcancer screening. Patients with a previous history of inflam-matory bowel disease, colorectal cancer, or polyposis syn-drome and those under surveillance for previous adenomasare not included. Most of the enrolled individuals are in theUW health system which covers Madison, WI, USA, and thesurrounding regions. A small number enter the programmefrom outside the system (geographically out-of-state or fromintrastate regions distant to the Madison area) because of itsreferral centre nature.

From this initial group, the individuals who receive generalmedical care from health providers outside the UW systemand whose only contact with the UW system was through asingle CTC examination were excluded from consideration(n030). Thus, the resultant group (n01,179) representedasymptomatic screening individuals enrolled in the CTCprogramme who also receive their general care within theUW health network. From this group, patients with a positiveexamination result (n0129) were then excluded. A positiveexamination consisted of at least one detected soft tissue polyp6 mm in size or greater. The resultant negative cohort con-sisted of 1,050 persons and represents the study population.This group consists of CRC-screened individuals with theresult of a normal CTC examination and the recommendationof repeat routine screening examination in 5 years. BecauseCTC examination positivity is set at a 6-mm polyp threshold,

individuals in this group may harbour diminutive polyps.Figure 1 outlines the study group determination.

CT colonography technique and polypectomy strategy

Standard CTcolonography technique, which was state-of-the-art for 2004–2005, was undertaken. It has been previouslydescribed in detail [9]. Bowel preparation consisted of a clearliquid diet, cathartic administration and a tagging regimen thatbegan 1 day before the scheduled examination. Catharsis wasundertaken by either a single 45-mL dose of sodium phos-phate, a double dose of magnesium citrate (300 mL per dose)or 1 L of polyethylene glycol. The tagging protocol includedthe administration of both 2.1% w/v barium sulphate(250 mL) and diatrizoate (60 mL).

Colonic distension was undertaken with an automatedcarbon dioxide delivery system (PROTOCO2L; Bracco) andlow-dose CTC was performed on either 8- or 16-section CT(LightSpeed Series; GE Medical Systems). The CTC tech-nique consisted of 1.25 mm collimation, 1-mm reconstructioninterval, 50–75 mAs and 120 kVp. All examinations werenetworked to a central reading area for post-processing andinterpretation (V3D colon, Viatronix).

The polypectomy strategy was in compliance with theCT colonography reporting and data system (C-RADS)guidelines [10]. Any polyp detected that was 6 mm in sizeor greater was referred for therapeutic colonoscopy. Forthose individuals with isolated (one or two) 6- to 9-mmpolyps, imaging surveillance was another option. Isolateddiminutive polyps were neither identified nor reported. Theroutine screening interval for a negative examination was5 years.

Determination of follow-up

The University of Wisconsin Health network is a verticallyintegrated system. It includes a tertiary referral hospital andlocal community-based hospital as well as local and regionalprimary care clinics in Madison and the surroundingregions. An electronic medical record (EMR) (EpicCareEMR, Epic) is integrated throughout the UW network.

The EMR for each patient was reviewed to assess for theinterval development of colorectal cancer. Records are regu-larly reviewed as part of quality assurance measures for theCTC programme. Length of follow-up was established by thenumber of days between the date of the CTC examination andthe date of the last comprehensive medical interactionrecorded in the EMR. A comprehensive medical interactionwas defined as a visit likely to record a history of colorectalcancer if present. These included any medical interaction(outpatient, emergency department, inpatient settings) wherea complete history and physical was completed as well asoffice visits for the specific purpose of health maintenance

Eur Radiol (2012) 22:1488–1494 1489

(preventive care) discussion. Anymedical interactions that didnot fall specifically into one of these two visit types werereviewed to assess whether they would potentially capturean interval history of cancer. For example, a telephoneencounter by a covering physician regarding an acuteillness would not be counted while a focused subspecialty visit(i.e. dermatology, cardiology) may or may not depending on thenature of the recorded progress note. For the group of personswho underwent interval colonoscopy or CT colonography forwhatever reason, the reports were reviewed for the presence ofcolorectal cancer. Any clinically significant neoplasms werealso recorded. Advanced adenomas were defined as adenomas10mm in size or greater or adenomas that harboured high-gradedysplasia or≥25% villous elements.

Statistical analysis

The crude cancer incidence rate was obtained by dividingthe number of people in the cohort with observed colorectalcancer events by the total time at risk, in person years. A95% confidence interval was calculated by assuming thenumbers of events to be Poisson-distributed and using anexact method [11].

Results

At a 6-mm threshold for a positive examination, individualswith a negative CTC result accounted for 89.1% (1,050/1,179)of the total CTC screening population. The mean age for thenegative CTC study population was 56.9 years of age with astandard deviation of 7.4 years. There was a female predom-inance of 54.0% (567/1,050). Table 1 shows the negative CTCdemographics including those of the two major subcohorts. Apositive family history was defined in accordance with theAmerican Cancer Society guidelines, which includes a first-degree relative with CRC before the age of 60 or two or morefirst-degree relatives with CRC of any age [5].

Of the 1,050 negative examinations, there were 39 exami-nations (3.7%) with a lack of follow-up within the EMRwhere the only record was the CTC examination for the in-network individual. This group was excluded from consider-ation. The remaining 1,011 patients had a documented aver-age follow up of 4.73 years (SD 1.15 years) with a total of4,786 patient years of follow-up. 595/1011 (58.9%) had atleast 5 years of documented follow-up, which corresponds tothe recommended interval for repeat examination for negativeCTC screening. Luminal colorectal evaluation by interval

Fig. 1 Flowchart for studycohort inclusion

1490 Eur Radiol (2012) 22:1488–1494

colonoscopy or CT colonography was performed in 368/1011(36.4%).

In the negative cohort, there was one incident colorectaladenocarcinoma that presented during the routine screeninginterval, leading to a rate of 0.1% (1/1,011). The crudeincidence rate was 0.2 cancers per 1,000 person-years(95% confidence interval 0.00529 to 1.16421 per 1,000person-years). With regard to the incident cancer, the patientpresented 35 months after a negative CTC examination withincreased rectal urgency and altered stool habits. Diagnosticwork-up revealed a Stage II (T3, N0, M0) low rectal ade-nocarcinoma. In retrospect, a questionable area of abnor-mality was present on the index CTC examination althoughthe area is difficult to assess due to adjacent redundanttissues and the rectal balloon (Fig. 2). The patient underwentneoadjuvant chemoradiation with complete surgical resec-tion. He remains disease free to date.

For those who underwent luminal examination, 14patients had clinically significant lesions (14/368; 3.8%).Eleven patients had an advanced adenoma. All advancedlesions met the criteria by large size (n011) alone; 5/11 werealso of tubulovillous histology. None harboured high-gradedysplasia. No subcentimetre advanced adenomas were seen.Overall, the per-patient advanced adenoma rate was 3.0%(11/368). The average time period between the index CTCand interval colonoscopy for this group was 5.1 years (SD0.5 years). There were five patients with six polyps ofunknown histology 6–9 mm in size detected at subsequentCT colonography at which the patient elected for imagingsurveillance.

The remaining three patients harboured significant peri-colonic gastrointestinal lesions including an appendicealgoblet cell carcinoid, an appendiceal mucinous tumour andan ileo-ceacal carcinoid with metastases to the liver. All

Table 1 Demographic charac-teristics of the patients

aFor comparisons betweenluminal and clinical follow-uponly cohorts

Entire studypopulation

Luminal examinationcohort (n0368)

Clinical follow-up onlycohort (n0643)

P valuea

Mean age (years) 56.9±7.4 56.3±6.7 57.3±7.6 0.036

Male:female ratio 483:567 174:194 285:358 0.394

Positive family hx 17 5 12 0.621

Fig. 2 A 69-year-old man who presented with an incident colorectalcancer 35 months after negative CTC screening (the only incidentcancer in this CTC-negative cohort). A lesion or polyp in this area isquestionably seen on the index examination. a 3D endoluminal imageat index CTC demonstrates an undulating longitudinal fold (arrow)adjacent to the rectal catheter tip. There is a questionable minimallyraised area (arrowheads) versus redundant tissues in this area. bTranslucency rendering demonstrates a homogeneous latticeworkappearance to the area arguing against a focal lesion (arrowheads,arrow) in the distal rectum. c 2D coronal reconstruction in theprone position shows the rectal fold (arrow) without definite lobulatedborders adjacent to the rectal catheter

b

Eur Radiol (2012) 22:1488–1494 1491

three patients presented symptomatically from 1 to 5 yearsfrom the index CTC. Surgical resection was performedduring the course of care in all cases. All patients are aliveat the time of review.

Discussion

Our study addresses the current paucity of data regardinglongitudinal colorectal outcomes in the CTC-screenedpopulation with a negative result and directly addresses thehypothetical concern that large numbers of incident cancersmay present clinically between negative CTC screeningexaminations. Such concern has arisen for two major reasons.

First, CTC-based screening differs from colonoscopywith regard to the management of subcentimetre polyps.The positivity threshold for the examination is set at 6 mmin which isolated diminutive lesions are neither identifiednor resected. The rationale for this alternative strategy is thatthese tiny polyps are very common (representing up to 76%of all polyps)[12] yet most never progress to cancer norharbour cancer (0.03% of diminutive polyps) [12]. Thefew potentially important polyps in this group can be iden-tified by demonstrating interval growth as they pass the 6-mm threshold on future routine screening examinations setat 5-year intervals. The benefit of this particular screeningmethodology is that it markedly decreases the number ofpolypectomies and procedure-related therapeutic complica-tions yet maintains programmatic advanced neoplasia yields[6]. Critics to this selective polypectomy approach suggestthat too many tiny cancers as well as high-risk polyps withthe potential to develop into incident cancers may be leftin vivo. Rex et al. reported that 5% of CTC-negativescreening patients may harbour high-risk adenoma findings[7]. However, it is important to realise that over 90% of thesecases will consist of patients with only three or more dimin-utive tubular adenomas, which are of uncertain but doubtfulimportance [13]. Only a very small fraction would harbourdiminutive advanced adenomas.

Second, flat polyps have been argued to be a clinicallysignificant lesion, more prevalent in the US population thanpreviously realised. Soetikno et al. reported that flat polypshad a greater association with cancer than polypoid lesionsregardless of size with a prevalence of nearly 6% in ascreening population and nearly 10% in a mixed population[8]. It has been argued that CTC has difficulty in detectingthese lesions given their minimally raised, plaque-like nature.However, in our experience, these lesions are often detectableat CTC, and their histology is significantly less aggressivethan polypoid and mass-like lesions [14].

The results of our study strongly argue against the clinicalrelevance of these debated concerns and suggest that the 5-year interval between negative CTC screening examinations is

a safe practice. In a large negative screening cohort, there wasonly a single incident symptomatic cancer that presented innearly 5 years of follow-up. This colorectal cancer outcomemirrors the results seen at colonoscopy screening – despite themarkedly different polyp harvesting strategies and the under-lying difference in technique. In a large cohort of 1,256individuals with a negative colonoscopic examination, Impe-riale et al. reported no cancers in the intervening 5 years [15].Given the substantial fraction (n0865; 35.5% of the originalgroup of 2,436) where neither colonoscopic nor follow-upinformation was available in the Imperiale series, the resultof one incident cancer in our study compares favourably withthe outcome seen in the colonoscopic study. Other smallnegative colonoscopic series have also demonstrated similarresults at 5-year follow-up [16–18].

The observed crude cancer incidence rate of our CTC-negative cohort (0.2 cancers/1,000 person years) is substan-tially lower than the rates observed in colonoscopic surveil-lance cohorts where rates range from 1.7 to 2.4 cancers per1,000 person years [19–21]. Colonoscopic surveillancegroups comprise individuals with positive index examinations(i.e. at least one adenoma identified and removed) who un-dergo clearing colonoscopy and are then followed for theinterval development of additional adenomas. The markeddifference in cancer incidence between these groups arguesthat the negative CTC population remains a low risk groupand is distinctly different from these colonoscopic sur-veillance cohorts. In other words, the situation where aCTC-negative screen may actually be similar to a colo-noscopic positive examination in surveillance is not likely. If aCTC-negative screen harboured clinically important adeno-mas that would have been detected if screened by colono-scopy (i.e. the debated significance of a diminutive adenoma),then the crude cancer incidence rate would presumably bemuch higher to echo the published colonoscopic surveillanceseries.

An argument may be raised that the lack of intervalluminal examination of all study participants in this seriesdiminishes the study results and conclusions. However, it isimportant to consider the following line of reasoning whythese study outcomes remain valid. The purpose of the studyis to provide evidence that the concerns of cancers and high-risk polyps left in situ at negative CTC screening are over-stated. If such lesions were present at the index examination,it would seem logical that these cancers would presentsymptomatically in the intervening years. Moreover, ourstudy suggests that it is highly likely that the incident cancerrate to include asymptomatic cancers that have not yetpresented clinically remains very low despite the lack offollow-up colonoscopy of the entire study population. In thesizeable cohort (n0368) that did undergo interval luminalevaluation, no cancers or adenomatous polyps with high-grade dysplasia were present. Thus, although it is possible

1492 Eur Radiol (2012) 22:1488–1494

that additional clinically silent cancers are present in theremainder of the population with clinical follow-up only(n0643), it is likely to be few to none in number as thissubgroup is statistically similar to the first group with regardto the major demographic determinants of adenoma andcancer prevalence (Table 1) [22, 23]. Between the twogroups, there was no statistical difference seen in the genderproportions or percentage with a positive family history. Astatistically significant difference was seen in the mean agesof the groups. However, such a difference of 1 year is ofdoubtful clinical relevance. Even in the unlikely circum-stance that several silent cancers are present in this group;the routine screen interval should detect these cancers at anearlier stage with improved future outcomes, arguingagainst the true clinical impact of the concerns outlined inthe previous paragraphs.

The advanced adenoma rate in this longitudinal studywas approximately 3% and all incident lesions were10 mm or larger in size. Despite the fact that theoreticallya few of these lesions were potentially present for severalyears, neither high-grade dysplasia nor asymptomaticcancers were identified. At first glance, the advanced neopla-sia incidence in this study appears slightly elevated in com-parison with colonoscopic series (reported rates of 1.3–1.4%at follow-up) [15–17]. However, this rate is in line with thesestudies if it is considered that the rate in the CTC-negativecohort represents the combination of advanced adenomaspresent in the diminutive group at the index examination aswell as ones that developed during the follow-up period.Large colonoscopic studies suggest that the prevalence ofadvanced neoplasia may be 0.87–1.7% in the diminutivegroup [7, 12]. Added to the percentage of advanced lesionsseen over a 5-year follow up after clearing colonoscopy, thenumber in the CTC-negative cohort is very similar to thecolonoscopic side. The lack of high-grade dysplasia (ormalignant foci) among these advanced lesions in our studyadds to the mounting evidence that transformation of theselesions occurs over the span of many years. These findingsreinforce the relative indolent nature of diminutive adenomas.However, the outcomes do point out that programmaticscreening compliance is required in order to detect theselesions. Screening at regular intervals would help to detectthese few important polyps in the diminutive group as theygrow to larger sizes before malignant transformation.

Regarding the limitations of this study, there were a smallnumber of individuals without follow-up information in theEMR. In addition, patients entered screening on referralfrom their primary care provider. Although they were gen-erally at average risk, the exact reasons leading to CTCscreening are unknown and selection bias cannot be com-pletely excluded. Finally, as discussed previously, it is alsopossible that tumours may not become clinically apparentduring the 5-year study window. However, this interval does

represent the recommended routine screening interval forCT colonography and as discussed previously, compliancewith routine screening would presumably identify theseclinically silent lesions, leading to earlier interventionand improved outcomes compared with later symptomaticpresentation.

In conclusion, the clinical presentation of an incidentcolorectal cancer is a rare event following a negative resultat CTC screening. The results of this large negative cohort atfollow-up suggest that the practice of setting a threshold of6 mm for examination positivity and a routine screen inter-val of 5 years is a safe practice despite the theoreticalconcerns that too many high-risk polyps and cancers are leftbehind. The low, but non-negligible incidence of advancedadenomas reinforces the necessity for programmatic com-pliance with regular routine screening.

Acknowledgements This work was supported in part by the researchgrant 1R01CA144835-01 from the National Cancer Institute.D.H.K is a consultant for Viatronix, co-founder of VirtuoCTC, and amember of the medical advisory board for Digital Artforms. P.J.P is aconsultant for Viatronix, Medicsight, Bracco, and Check-cap, and is aco-founder of VirtuoCTC.

References

1. Pickhardt PJ, Choi JR, Hwang I et al (2003) Computed tomo-graphic virtual colonoscopy to screen for colorectal neoplasia inasymptomatic adults. N Engl J Med 349:2191–2200

2. Johnson CD, Chen MH, Toledano AYet al (2008) Accuracy of CTcolonography for detection of large adenomas and cancers. N EnglJ Med 359:1207–1217

3. Graser A, Stieber P, Nagel D et al (2009) Comparison of CTcolonography, colonoscopy, sigmoidoscopy, and fecal occult bloodtests for the detection of advanced adenoma in an average riskpopulation. Gut 58:241–248

4. Regge D, Laudi C, Galatola G et al (2009) Diagnostic accuracy ofcomputed tomographic colonography for the detection of ad-vanced neoplasia in individuals at increased risk of colorectalcancer. JAMA 301:2453–2461

5. Levin B, Lieberman DA, McFarland B et al (2008) Screening andsurveillance for the early detection of colorectal cancer and adeno-matous polyps, 2008: a joint guideline from the American CancerSociety, the US Multi-Society Task Force on Colorectal Cancer, andthe American College of Radiology. CA Cancer J Clin 58:130–160

6. Kim DH, Pickhardt PJ, Taylor AJ et al (2007) CT colonographyversus colonoscopy for the detection of advanced neoplasia. NEngl J Med 357:1403–1412

7. Rex DK, Overhiser AJ, Chen SC, Cummings OW, Ulbright TM(2009) Estimation of impact of american college of radiologyrecommendations on CT colonography reporting for resection ofhigh-risk adenoma findings. Am J Gastroenterol 104:149–153

8. Soetikno RM, Kaltenbach T, Rouse RV et al (2008) Prevalence ofnonpolypoid (flat and depressed) colorectal neoplasms in asymp-tomatic and symptomatic adults. JAMA 299:1027–1035

9. Pickhardt PJ, Taylor AJ, Kim DH, Reichelderfer M, Gopal DV,Pfau PR (2006) Screening for colorectal neoplasia with CT colo-nography: initial experience from the 1st year of coverage by third-party payers. Radiology 241:417–425

Eur Radiol (2012) 22:1488–1494 1493

10. Zalis ME, Barish MA, Choi JR et al (2005) CT colonographyreporting and data system: a consensus proposal. Radiology236:3–9

11. Ahlbom A (1993) Biostatistics for epidemiologists. CRC, BocaRaton

12. Lieberman D, Moravec M, Holub J, Michaels L, Eisen G (2008)Polyp size and advanced histology in patients undergoing colonoscopyscreening: implications for CT colonography. Gastroenterology135:1100–1105

13. Pickhardt PJ, Kim DH (2009) Advanced vs. “High-Risk” adenomas:identifying the relevant target for CT colonography screening. Am JGastroenterol 104:1599–1600

14. Pickhardt PJ, Kim DH, Robbins JB (2010) Flat (Nonpolypoid)colorectal lesions identified at CT colonography in a US screeningpopulation. Acad Radiol 17:784–790

15. Imperiale TF, Glowinski EA, Lin-Cooper C, Larkin GN, Rogge JD,Ransohoff DF (2008) Five-year risk of colorectal neoplasia afternegative screening colonoscopy. N Engl J Med 359:1218–1224

16. Leung WK, Lau JYW, Suen BY et al (2009) Repeat-screeningcolonoscopy 5 years after normal baseline-screening colonoscopyin average-risk Chinese: a prospective study. Am J Gastroenterol104:2028–2034

17. Rex DK, Cummings OW, Helper DJ et al (1996) 5-year incidenceof adenomas after negative colonoscopy in asymptomatic average-risk persons. Gastroenterology 111:1178–1181

18. Neugut AI, Jacobson JS, Ahsan H et al (1995) Incidence andrecurrence rates of colorectal adenomas—a prospective-study.Gastroenterology 108:402–408

19. Robertson DJ, Greenberg ER, Beach M et al (2005) Colorectalcancer in patients under close colonoscopic surveillance. Gastro-enterology 129:34–41

20. Pabby A, Schoen RE, Weissfeld JL et al (2005) Analysis ofcolorectal cancer occurrence during surveillance colonoscopy inthe dietary polyp prevention trial. Gastrointest Endosc 61:385–391

21. Alberts DS, Martinez ME, Roe DJ et al (2000) Lack of effectof a high-fiber cereal supplement on the recurrence of colorectaladenomas. N Engl J Med 342:1156–1162

22. Arminski TC, McLean DW (1964) Incidence and distribution ofadenomatous polyps of the colon and rectum based on 1,000autopsy examinations. Dis Colon Rectum 7:249–261

23. Gaglia P, Atkin WS, Whitelaw S et al (1995) Variables asso-ciated with the risk of colorectal adenomas in asymptomaticpatients with a family history of colorectal-cancer. Gut 36:385–390

1494 Eur Radiol (2012) 22:1488–1494