Virtual Colonscopy for Primary Screening

MINERVA CHIR 2005;60:139-50

A Virtual colonscopy for primaryC screening IThe future is now D P. J. PICKHARDT 1, 2 E Virtual colonoscopy (VC) is a minimally invasive ®1Department of Radiology tool that utilizes modern CT technology for col- University of Wisconsin Medical School orectal evaluation. Since its inception in 1994, VC Madison, WI, USA has continued to rapidly evolve and improve M 2Department of Radiology as a diagnostic screening tool. Early success Uniformed Services University using primary two-dimensional (2D) detection T of the Health Sciences in polyp-rich cohorts was followed by disap- Bethesda, MD, USA pointing results in low prevalence populations. Subsequent introduction of the three-dimensional (3D) endoluminal display for primaryA H polyp detection and oral contrast tagging has transformed VC into an effective primary screen- irtual colonoscopy (VC), also referred to ing tool. This state-of-the-art VC technique has Vas CT colonography, is a minimally inva- already proven to be a viable enterpriseV when sive test for the detection of colorectal polyps combined with existing optical colonoscopy and masses.G This technique, which combines practice. More widespread implementation of two-dimensionalI (2D) and three-dimensional VC screening faces multiple challenges,R but these (3D) CT displays, has been rapidly evolving are all greatly overshadowed by the immediate 1 need for increased participation in effective col- since its inception in 1994. Some of the ear- orectal screening. Given its relatively noninva- lier VC trials involving polyp-rich cohorts sive nature and the wide Eavailability of CT, VCRdemonstrated very encouraging results using holds significant potential for addressing a very a primary 2D approach to polyp detection,2, 3 important yet preventable public health con- but the initial attempts to study low preva- cern. This review will cover current VC tech- Y lence populations were rather disappointing.4- nique, compare the existing multi-center VC tri- N 6 als, discuss issues related to primary VC screen- However, subsequent improvements in VC ing, and briefly Iupdate the progress of our VC technique, particularly the use of the 3D endo- screening program. P luminal display for primary polyp detection Key words: Virtual colonoscopy - Colon diseases and oral contrast for tagging of residual fluid - Rectum diseases - Computer tomography. and stool, have transformed VC into an effective primary screening tool. These advances The opinionsM and assertions containedO herein are the pri- vate views of the authors and are not to be construed as offi- culminated in a large multi-center VC screen- cial or as reflecting the views of the Departments of the Navy ing trial that showed comparable performance or Defense. to optical colonoscopy (OC).7 Clinical imple- C mentation of this proven method for VC Address reprint requests to: P. J. Pickhardt, MD, Department screening has already been shown to be a of Radiology, University of Wisconsin Medical School, E3/311 Clinical Science Center, 600 Highland Ave, Madison, WI 53792- viable enterprise, particularly when covered by 3252. E-mail: [email protected] third-party payers and combined with an exist-

Vol. 60, N. 3 MINERVA CHIRURGICA 139 PICKHARDT VIRTUAL COLONSCOPY FOR PRIMARY SCREENING

TABLE I.—Potential challenges to clinical implementa- Colonic preparation tion of VC screening. Robust colonic preparation is important — Reimbursemnt of VC screening from third-party payers for accurate polyp detection.7, 10 Not only (U.S.) — Development of an acceptable diagnostic screening must the fecal matter be adequately removed, but oral contrast material shouldA be used for algorithm — Identifying the appropriate group of patients for pri- tagging any retained solid or liquid material. mary VC screening Patient preparation usually begins the day — Establishing an effective relationship with gastroente- before the examination. We have had great rology and colorectal surgery C — Establishing practice guidelines and program accredi- success with a relatively simple low-volume tation cathartic preparationI (less than 400 mL) that — Demonstrating cost effectiveness of VC screening combines 3 basic components, all of which — Dedicated training of enough radiologists and techno- are important for complete success: sodium logists phosphate for catharsis, dilute 2% barium for — Educating patients and referring physicians on the role D for VC in screening solid stool tagging, and an ionic water-solu- — Continued need for colon purgation (and development ble contrast agent for fluid opacification.10 In of non-cathartic alternatives) addition to the actual prep components, the — Issue of exposure to ionizing radiation (albeit low dose patient maintainsE a clear-liquid diet with lib- in adults) ® — Assessing the net impact of extracolonic CT findings eral hydration throughout the day. Through — Competing with other screening tools (new and old) continued testing, we have refined this prep — Flat colorectal lesions considerably compared with that used in the multi-centerM trial.7 Now, only single doses of each component Tare required the evening ing OC program.8 The dire need for increased before the examination (45 mL sodium phos- colorectal screening is quite apparent, as far phate, 250 mL 2% barium, and 60 mL gas- too many people are dying from a largely pre-A troview or gastrografin), which greatly sim- H 11 ventable disease.9 Although widespread imple- plifies the patient instructions. For patients mentation of VC screening faces multiple chal- with known or suspected renal or cardiac lenges (Table I), these are clearly overshad-V insufficiency, magnesium citrate is substitut- owed by the need for increased screening. ed forG the sodium phosphate. This review will cover the basic essentials of Less vigorous preparations for VC are cur- current VC technique, compare the 3 major rentlyI under investigation and have shown R 12 multi-center VC trials to date, address some of preliminary success in high-risk groups. The the relevant issues related to primary VC prepless designation sometimes applied to this approach is a misnomer, since a bowel screening, and provide an update on the sta-R E preparation is invariably employed. The terms tus of our VC screening program. “non-cathartic” or “minimal” prep are closer to the truth, although most would agree that Y the use of gastrografin provokes at least a VC Ntechnique mild cathartic effect. Regardless, important I trade-offs with this approach include an The basic concept behind VC isP rather sim- expected drop in accuracy in low prevalence ple: by imaging a properly cleaned and dis- populations and inability for same-day tended colon with a modern (multi-detec- polypectomy.13 Although non-cathartic preps tor) CTM scanner, clinically significant col- for VC may ultimately increase overall com- orectal polyps can be readilyO identified. pliance, I believe it should be primarily Having said that, there has been consider- reserved for patients who are unwilling to able variability in the specific techniques used undergo proper cleansing, assuming they for performing VC. ForC the purpose of this understand the consequences. Furthermore, report, emphasis will be placed on the VC this approach must first be validated in a methods that have proven most successful multi-center trial evaluating an asymptomatic to date. average-risk population. In my opinion, it is

140 MINERVA CHIRURGICA Giugno 2005 VIRTUAL COLONSCOPY FOR PRIMARY SCREENING PICKHARDT

preferable to use a bowel preparation method that allows for the possibility of same-day polypectomy to avoid the need for a second prep. Our patients greatly value this one-stop shop approach to colorectal screening. A Colonic distention Gaseous distention of the colon, like prop- C er cleansing, is also critical for diagnostic success. Distention may be achieved with either I room air or carbon dioxide. The rate and degree of insufflation can be controlled by the patient, controlled by the technologist/physi- D cian, or automated. The adequacy of colonic distention prior to scanning is usually gauged on the CT scout view (Figure 1). The 2 best E (and safest) techniques for consistent colonic ® distention are patient-controlled room air insufflation and automated carbon dioxide delivery (PROTOCO2L, E-Z-EM). Perforations FigureM 1.—Scout image from CT scan performed as part of are extremely rare when these techniques VC study in 53-year-old woman undergoing colorectal are employed and are essentially unheard screening. Excellent gaseousT distention such as this is relia- bly obtained from the automated carbon dioxide delivery of in the setting of asymptomatic VC screen- system (PROTOCO2L, E-Z-EM). The sigmoid colon, howe- ing. We recently compared patient-controlled ver, often requires extra attention and its adequacy for room air insufflation versus automated car-A distention is bestH evaluated directly on the axial 2D images. bon dioxide delivery for VC screening in over 200 patients and determined that the now offer these patients same-day, unse- latter approach resulted in slightly improvedV datedG flexible sigmoidoscopy to complete colonic distention and decreased post-pro- their screening evaluation (again, a one-stop cedure discomfort (Shinners TJ, Pickhardt shopI offering). PJ, Taylor AJ, Jones DA, OlsenR CH. Colonic distention and patient comfort at screening CT colonography: prospective comparison CT scanner requirements of patient-controlled roomE air insufflationRWhile it is true that a multi-detector CT versus automated carbon dioxide delivery. scanner is necessary for acceptable 3D image Submitted). Furthermore, the automated car- quality at VC, a four-channel CT scanner will bon dioxide method was clearly preferredY generally suffice. This is due in part to the fact over patient-controlledN room air insufflation that the gas-filled colon is a relatively static by the CT technologistsI obtaining the stud- and forgiving structure, but also because the ies. As a result of these findings,P automated target lesions are sufficiently large (i.e., polyps carbon dioxide delivery has now become measuring 5-6 mm or greater in size) and do our front-line method for colonic distention. not require exquisite spatial resolution for In a small number of patients undergoing detection. Although we now perform all VC primaryM VC evaluation (2%O or less, in our exams on eight- and 16-channel CT scanners experience), the study may be incomplete (with 1.25 mm collimation and 1-mm recon- due to focal collapse of a colonic segment on struction interval), a 4×2.5 mm detector con- both supine and proneC views. Because the figuration remains adequate for successful great majority of nondiagnostic segments examination.7, 8 Given the nature of the soft involve the sigmoid colon, typically as a tissue-air interface, the CT technique for VC result of extensive diverticular disease, we entails significantly lower radiation doses

Vol. 60, N. 3 MINERVA CHIRURGICA 141 PICKHARDT VIRTUAL COLONSCOPY FOR PRIMARY SCREENING

compared with standard abdominal CT stud- This interpretation approach is best described ies. The clear benefit of effective colorectal as a biphasic examination. The need for high- screening outweighs the small theoretical risk level 3D-2D correlation requires that CT- from the low-dose radiation exposure in the trained radiologists interpret these examina- adult population.14 tions. Although there have been significant recent improvements in theA 3D rendering VC software: 2D vs 3D polyp detection capabilities of many VC software systems, it is unfortunate that very few systems currently The choice of VC software for interpreta- allow for accurate and time-efficientC primary tion is an absolutely critical factor, greatly 3D navigation.15 outweighing the small differences that exist The major advantageI with using the 3D among different multi-detector CT scanners. endoluminal view for primary polyp detec- Of paramount importance is whether or not tion is that the visual search pattern is much effective and time-efficient 3D evaluation is simpler comparedD with 2D.15, 16 Polyp con- even possible with a given VC software sys- spicuity is greatly enhanced and distin- tem. Primary 2D evaluation for polyps con- guishing polyps from folds is a much easier sists of scrolling through the hundreds of 2D task on 3D.E In comparison, the eye strain CT images in a manual lumen tracking mode. and fatigue associated® with primary 2D eval- In my opinion, there are several reasons why uation becomes quite evident when faced this 2D approach has unfortunately become with interpreting multiple cases in succes- ingrained as the leading search method: 1) sion.M Approximately 20 000 axial 2D images high-quality 2D displays were available when would need to be viewedT for each advanced VC was first emerging as a nascent technique adenoma that will be encountered in a typ- in the mid-late 1990s; 2) a pre-existing com- ical screening population. Furthermore, our fort level for interpreting cross-sectional CT experience shows that flat lesions do not images made 2D evaluation for polyps a log-A represent aH significant drawback for VC ical extension for abdominal CT radiologists; screening in typical Western populations 3) effective primary 3D endoluminal evalua- when state-of-the-art 3D imaging is com- tion was simply not feasible during Vearly VC bined with 2D imaging.17 Our investigations development. have alsoG shown that linear polyp measure- Primary 2D polyp detection served well mentI on the 3D endoluminal view is signif- for early proof of concept VCR studies that icantly more accurate than 2D axial, coro- evaluated polyp rich cohorts 2, 3 but per- nal, or sagittal measurements, both in vitro formed poorly when applied to low preva- and in vivo.18 This is due to both 2D win- lence populations, as discussedE below.4-6Rdowing effects and the fact that the long axis Fortunately, effective primary 3D polyp detec- of a polyp is easily measured on 3D but tion became available (V3D Colon, Viatronix) rarely aligns directly with a standard orthog- before we embarked upon our multi-centerY onal 2D view. Optimizing the linear 2D size screening trial (FigureN 2).7, 15 The results from by obtaining all 3 orthogonal measurements our trial (also discussedI below) demonstrat- minimizes but does not eliminate 2D under- ed that VC with emphasis on 3DP detection sizing of polyps on 2D. compared favorably with OC.7 When viewed There are several additional 3D tools and side-by-side, it becomes fairly obvious why features that can further benefit the radiolo- 3D polypM detection represents an inevitable gist’s interpretation. Translucency rendering successor to 2D polyp detection.O In practice, is a tool that allows for rapid distinction this transition from 2D to 3D polyp detec- between soft tissue polyps and adherent tion represents a fundamental reversal in the tagged stool on the 3D view, reducing the approach to VC interpretation.C The 2D images need for time-consuming 2D correlation.19 A now primarily serve as source images that missed patch tool allows for tracking and are vital for confirming 3D findings, but also visualization of mucosal surfaces that are not have a secondary role in polyp detection. seen during standard bi-directional navigation

142 MINERVA CHIRURGICA Giugno 2005 VIRTUAL COLONSCOPY FOR PRIMARY SCREENING PICKHARDT

A C I D E ® M T A H V G R I E R

Figure 2.—Ten-mm tubulovillous adenoma in an asymptomatic 61-year-old man undergoing VC screening. A) Colonic map with centerline for 3D navigation (black line)Y generated from VC software (V3D Colon, Viatronix). The automated nature of this step increasesN reading efficiency for the radiologist. The map also provides an effective means for communicating the location of detected polyps. B-D) Three-dimensional endoluminal view from VC (A), axial 2D CT image from VC (B), and digitalI photograph from OC (C) all show the same polyp, which was located near the rectosigmoid junction. The conventional study was performedP within 2 h of the virtual study; all pertinent images from VC were shared with the endoscopist prior to polypectomy. along theM automated centerline.O Electronic communicating findings to endoscopists fluid cleansing subtracts opacified luminal (Figure 2). Bookmarking features provide fluid and increases mucosal surface visual- precise localization of polyps for either sub- ization, but also introducesC artifacts that can sequent VC surveillance or OC polypecto- be distracting (which is why we currently do my. The automated centerline for luminal not employ it).10 The colon map provides an navigation confers a significant time savings overview of colonic anatomy and is useful for compared with systems where the flight path

Vol. 60, N. 3 MINERVA CHIRURGICA 143 PICKHARDT VIRTUAL COLONSCOPY FOR PRIMARY SCREENING

must be manually determined by the radiol- To properly assess a new colorectal screen- ogist (Figure 2). ing tool, it would seem logical to evaluate a true screening population. However, neither the Rockey trial nor Cotton trial studied an Comparison of the multi-center VC asymptomatic screening population,A where- trials as all adults in the Pickhardt trial were asymptomatic. Significant differences in VC perfor- There have been 3 large multi-center VC mance might well be expected between symp- trials that have been published to date: tomatic and asymptomaticC cohorts. Another Pickhardt et al.,7 Cotton et al.,5 and Rockey et methodological drawbackI in the Cotton and al.6 The results were strikingly different, with Rockey trials was the small patient popula- a by-patient sensitivity of 94% for adenomas at tion relative to the number of study sites, result- the 10-mm threshold in the Pickhardt trial, ing in an average of less than 50 patients per compared with sensitivities of 55% and 59% in site, only a fewD of whom would be expected the Cotton and Rockey trials, respectively (the to harbor an adenoma 10 mm or greater. This figures for the Cotton and Rockey trials are in sampling bias is further compounded by the terms of all polyps – adenomas were not con- fact that Estudy radiologists did not receive sidered separately). There was less of a dis- feedback to allow for® performance improve- parity between by-patient specificities, but this ment during the trial. In comparison, the is of course of secondary importance com- Pickhardt trial was twice the size of the Rockey pared with sensitivity. Of particular impor- andM Cotton trials, involved an average of more tance to note is that these studies actually tran- than 400 studies perT site, and allowed for spired in reverse order of their publication, ongoing performance feedback, all of which with Rockey et al. and Cotton et al. preceding enable a better indication of how this new the more recent and larger study by Pickhardt screening tool might perform in actual practice. et al. by more than 2 years. With this in mind,A The resultsH of the studies by Cotton and one can begin to appreciate that the corre- Rockey would have been much more valu- sponding results could largely reflect the ongo- able had they been published in their prop- ing technical advances in this rapidly-evolvingV er chronological and evolutionary order. The field. Furthermore, it stands to reason that a confusionG created by their delayed publica- radiologist-led study (Pickhardt trial) would tionI has led to the misconception that VC be more familiar and facile withR advances in performance is somehow going backwards. CT imaging than gastroenterologist-led studies Nonetheless, it serves as another reminder such as the Rockey and Cotton trials. that specific techniques, particularly 3D polyp Although there are manyE potential causesRdetection and contrast tagging, truly matter for the variable performance seen in these tri- and that further advances in this important als, 2 factors rise above all others. Most impor- area of radiologic imaging are best handled tantly, these studies collectively demonstrateY by radiologists. Of concern for the future is that primary 3D polypN detection, used only in that ongoing or upcoming VC trials may sim- the Pickhardt trial,I is vastly superior to 2D detec- ply try to co-opt the banner of 3D emphasis tion in low-prevalence populations.P The second but actually employ the typical VC software major factor is that tagging of residual colonic system that does not allow for effective and stool and fluid with oral contrast, which was efficient primary 3D evaluation. also usedM only in the Pickhardt trial, is now recognized as a critical componentO of VC success.20 Regrettably, the individual contributions Issues related to primary VC screening to the decrease in accuracy resulting from the Developing a diagnostic algorithm lack of 3D polyp detectionC and contrast tagging in the Cotton and Rockey trials cannot be For patients undergoing colorectal screen- ascertained. Regardless, both played a critical ing with VC, the largest detected lesion large- role in the success of the Pickhardt trial. ly determines the next appropriate step. A

144 MINERVA CHIRURGICA Giugno 2005 VIRTUAL COLONSCOPY FOR PRIMARY SCREENING PICKHARDT

diagnostic algorithm based on polyp size clinical management of medium-sized polyps stratifies patients into categories such as (6-9 mm) detected at VC. However, it is impor- immediate OC for polypectomy, short-term tant to keep in mind that this issue is clearly sec- VC surveillance, and routine follow-up.21 ondary to increased screen detection of large Adoption of reasonable polyp size thresh- polyps in a greater percentage of adults. olds and follow-up intervals will be critical to Fueling this potential controversyA over the the ultimate success of VC screening.14, 22, 23 management of medium-sized lesions is the For patients with medium-sized (6-9 mm) relative dearth of information about the natural colorectal polyps detected at VC, we offer history of colorectal polyps.C There is, howev- the option of immediate polypectomy at OC er, some encouraging news that is often over- versus noninvasive VC surveillance. We looked: the existingI data on polyp growth believe this represents a logical and clinical- rates, although somewhat limited, clearly sup- ly sound strategy. The rationale behind this port the validity of noninvasive surveillance paradigm shift lies in both the indolent nature of small sub-cmD polyps. Several longitudinal of sub-centimeter polyps and the noninvasive OC studies have been published that followed method of detection. Once a patient has com- unresected sub-cm polyps over time, all of mitted to primary OC for evaluation and has which haveE demonstrated the indolent nature accepted the risks of this more invasive proof these small colorectal® polyps. Hofstad et al. cedure, removal of all nondiminutive polyps reported on the largest and most meaningful of is expected. The situation is quite different for these studies, which followed unresected VC screening because the exclusion of large polypsM up to 9 mm with serial OC examinations polyps (≥ 10 mm) places the patient in a very over a three-year period.T29, 30 After the first year low risk category and the risk/benefit ratio for of surveillance, only one (0.5%) of 189 sub- subsequent OC is now less favorable.24, 25 cm polyps crossed the 10-mm size threshold.29 There is general agreement that immediate At 3 years, the majority of polyps either polypectomy is indicated for large polypsA remained stableH or regressed in size, but more detected at VC screening. In our experience, importantly, 5-9 mm polyps actually showed an this situation occurs approximately once overall tendency for regression.30 The every 20 cases; although most of theseV large researchers concluded that follow-up of unre- lesions represent advanced adenomas, very sectedG colorectal polyps up to 9 mm is a safe few will harbor frank malignancy. One could practice,I which closely parallels our current argue whether a 10-mm thresholdR is too low, approach for noninvasive VC surveillance. In since it is estimated that fewer than 10% of an earlier study, this same group showed that polyps ≥ 10 mm will actually develop into diminutive tubular adenomas rarely develop in cancer at 10 years.26 Nonetheless,E the risks ofRto advanced lesions in the short term.31 In a undergoing polypectomy by OC are likely smaller series, Bersentes et al. showed no sig- outweighed by the malignant potential of nificant change in the size of sub-cm polyps at these larger polyps. With regard to subcen-Y two-year follow-up, with only a minority of timeter colorectal Npolyps, there also appears lesions growing more than 1 mm in diame- to be general Iconsensus that diminutive ter.32 Finally, review of data from the National lesions (≤ 5 mm) are of no practicalP clinical Polyp Study concluded that the high observed significance.7, 27 Only a fraction of these adenoma detection rates at surveillance diminutive polyps are neoplastic, and of colonoscopy and the low observed colorectal these, Mless than 1% are histologically cancer incidence seen in this large study could advanced and essentiallyO none are malig- only be explained by assuming a high inci- nant.27, 28 As such, patients without any polyps dence rate of adenoma regression.33 measuring greater than 5 mm at VC are con- Establishing appropriate follow-up intervals sidered to have an essentiallyC normal study in for patients with sub-centimeter polyps or no our screening program. polyps detected at VC represents the other In my opinion, much of the controversy sur- major challenge for developing an agreeable rounding VC screening will center around the diagnostic screening algorithm. The routine

Vol. 60, N. 3 MINERVA CHIRURGICA 145 PICKHARDT VIRTUAL COLONSCOPY FOR PRIMARY SCREENING

follow-up interval following a negative VC that finding polyps with VC was feasible study (i.e., no polyps detected > 5 mm) could when carried out in polyp-rich cohorts. safely be set at 5 years to coincide with the However, the ideal population for primary accepted intervals for sigmoidoscopy and the VC screening consists of asymptomatic barium enema. As more experience and data adults.21 For one, these patientsA are least like- are gathered, this interval would likely expand ly to require subsequent OC for polypectomy, towards the ten-year level that is currently whereas consideration for primary OC eval- accepted following negative OC examination. uation would seem more prudent in symp- Although three-year follow-up of unresected tomatic patients or thoseC with a high a priori medium-sized polyps (6-9 mm) was shown risk for neoplasia, givenI its therapeutic capa- to be a reasonable approach in the Hofstad tri- bility. This risk stratification approach would al, more conservative parameters may be minimize the number of patients requiring employed during the early implementation both examinations. Given the short supply of phase of VC screening. The Working Group on gastroenterologists,D I believe that OC needs Virtual Colonoscopy has recently issued pre- to be viewed as a limited resource that is bet- liminary guidelines for VC screening that call ter utilized as a therapeutic procedure for for a 5-10 year interval for a negative VC exam- removingE significant polyps, rather than being ination and 3-year follow up for unresected used for negative diagnostic® exams. The argu- medium-sized polyps.34 It remains to be seen ment could be made that the known risks of how swiftly these guidelines will be adopted perforation and clinically significant bleed- by real-world practitioners. ingM associated with OC polypectomy, occur- Another potential inclusion with the diag- ring in about 0.2% Tand 1-2% of cases, respec- nostic algorithm would be to stratify VC find- tively,24, 25 outweigh the minuscule cancer ings according to diagnostic confidence of risk posed by sub-cm polyps. Seemingly rare the interpreting radiologist. We have found risks of a test are amplified when applied to that increased reader confidence for an indi-A a large numberH of people in an otherwise vidual lesion detected at VC correlates with healthy screening population. In particular, a significantly increased likelihood Vthat: 1) a perforation at OC may be particularly dev- matched polyp will be found at OC and, 2) astatingG (both physically and emotionally) that this matched polyp will be neoplastic.35 when it occurs in a healthy 50-year-old adult In summary, the initial debate over how to - primumI non nocere. Therefore, reserving manage medium-sized polyps Rdetected at VC OC for removal of the more important, large screening will no doubt be spirited at times. colorectal polyps found by VC in asympto- However, regardless of how we decide toRmatic adults not only helps to preserve a handle sub-centimeter polypsE detected at VC valuable resource, it is also a sensible strate- screening, the emphasis should be placed gy from a risk/benefit standpoint. on the detection of larger polyps, particular-Y ly in patients whoN otherwise may not have Extracolonic CT findings identified at VC been screened at all. In my opinion, con- screening centrating effortsI more on the detection and removal of less common but moreP dangerous Extracolonic evaluation from the 2D CT lesions in the majority of the screening pop- images at VC represents a double-edged ulation would represent a far more efficient, sword. On one hand, potential benefits cost-effective,M and efficaciousO strategy than include personal reassurance for the vast the current practice of harvesting many small majority in whom nothing ominous is found polyps in the minority. and, in a small minority, discovery of an unsuspected but clinically significant process at an early, pre-symptomatic stage. On the Defining the appropriateC patient population other hand, potential drawbacks include The early VC studies essentially provided unnecessary anxiety and added costs related necessary proof of concept, demonstrating to additional work-up for findings that ulti-

146 MINERVA CHIRURGICA Giugno 2005 VIRTUAL COLONSCOPY FOR PRIMARY SCREENING PICKHARDT

mately prove to be of no consequence. orectal screening is more about cancer pre- Unfortunately, most studies to date on this vention than cancer detection.7 Regardless topic have involved symptomatic or high- of whether one views extracolonic evalua- risk individuals, for which extracolonic find- tion resulting from VC screening as a net ben- ings are more frequent.36-39 In contrast, poten- efit or liability, it is an unavoidableA respon- tially significant extracolonic findings are sibility that must be handled with care by the encountered in asymptomatic adults at a interpreting radiologist. much lower frequency.40 It must be empha- sized that the lack of IV contrast and the low Collaboration with endoscopistsC (gastroen- dose technique employed for VC limit the terology and colorectal surgery) evaluation of CT findings outside of the colon. I For radiologists performing VC screening, a It will be very important for radiologists, careful log of potentially important extra- gastroenterologists, and colorectal surgeons colonic findings should be kept and period- to work togetherD closely as colorectal screen- ically checked to confirm resolution. ing evolves. VC screening, if properly imple- The vast majority of extracolonic findings mented, will result in a win for all, including at VC are of essentially no clinical signifi- society asE a whole. The potential added ben- cance and include things such as uncompli- efit of VC screening® is obvious, since any cated renal or hepatic cysts, mild arterial vas- reasonable increase in effective colorectal cular calcification, hernias (particularly hiatal screening should lead to a net reduction in and inguinal), and benign skeletal findings cancerM incidence and mortality. For radiolo- (e.g. degenerative changes). Uncomplicated gists, there is nowT a viable and exciting cholelithiasis and nephrolithiasis may be opportunity to make a substantive contribu- encountered in 5-10% of cases, but rarely tion to colorectal cancer screening. For gas- requires further work up. Published studies troenterologists and colorectal surgeons, VC have tended to report on the frequencyA of screening shouldH be viewed not as a threat extracolonic findings at VC in terms of mod- but rather as a means to enhance their clini- erate importance and high importance, but cal practices, since it serves as a complement this practice greatly overstates the frequencyV and not a replacement for the existing screen- of truly significant findings, since even most ing options.G41 VC could be initially targeted to in the highly important category ultimately thoseI who are either reluctant to undergo prove to be of no consequence.R 36-40 To OC or are at significantly increased risk for a address this issue, we report such findings complication.14 In our experience, the net to be of potential importance to underscore effect of this approach has been an increased both the need for furtherE evaluation and theRfrequency of therapeutic OC, relatively less reasonable likelihood for a good outcome.7, time spent on negative (nontherapeutic) 40 Extracolonic findings that may require fur- exams, and an overall increase in the total ther evaluation include complex-appearingY number of OC exams performed. One would adnexal lesions in Npost-menopausal women, assume that the level of gratification resulting indeterminant Irenal lesions, indeterminant from endoscopic removal of an advanced hepatic lesions, and lymphadenopathy.P Other neoplasm would exceed that of a negative occasional findings include abdominal aortic invasive examination. aneurysms, adrenal adenomas, pulmonary By providing a comprehensive colorectal nodules,M and a variety of incidental congen- screening service, it is possible to provide a ital variants, In our experience,O unsuspect- one-stop shop experience for patients. When ed extracolonic malignancy is seen in approx- the radiologist reads the VC examination on- imately 1 case per 200 patients screened.40 line, a fasting patient can then undergo Interestingly, VC willC typically uncover more polypectomy at OC without the need for a extracolonic malignancies than colon can- second colonic preparation. This option is cers in a typical asymptomatic screening pop- highly valued by many patients but it requires ulation, which underscores the fact that col- dedication from the radiologist to rapidly inter-

Vol. 60, N. 3 MINERVA CHIRURGICA 147 PICKHARDT VIRTUAL COLONSCOPY FOR PRIMARY SCREENING

pret the VC study. A commitment from the Current status of the VC screening endoscopist is also required to provide this program at UW same-day service. This cooperative arrange- ment functions as a two-way street, since In April 2004, the VC program at the same-day VC add-ons for incomplete OC also University of Wisconsin becameA the first in occur. As described below, this is the approach the US to receive standard third-party reim- we have employed for the VC screening pro- bursement for VC screening.8, 43 One year lat- gram at the University of Wisconsin. er, it unfortunately remains the only program The specific impact of VC screening on col- in the US to enjoy thisC status. The reasons orectal surgery practice promises to be very for delay in VC coverageI at other centers are positive. In addition to detecting more multifactorial, but the use of unproven or advanced adenomas, an increase in effective poorly performing VC techniques is a major colorectal screening should also result in the issue. For the past 6 months, we have been detection of more cancers at a surgically curable routinely performingD about 10 screening stud- stage. It has been shown that detection of ies per day between 7 and 10 am. All studies asymptomatic colorectal cancer offers a better are read on-line to allow for same day prognosis and is more cost effective than wait- polypectomy,E if needed. Approximately 5- ing for symptomatic detection.42 VC can provide 10% of patients ®go on to same-day exquisite preoperative planning, with precise colonoscopy, the vast majority of which are tumor localization and also evaluation of extra- performed on the same day. Put another way, colonic structures. Beyond mucosal-based overM 90% of our patients that undergo VC lesions, VC can provide much more complete screening avoid theT need for OC. Without characterization on submucosal lesions, any organized marketing or advertising cam- because CT evaluates the entire bowel wall.16 paign, patient demand for VC has remained VC has also proven useful for localization and high, with hundreds of patients now sched- characterization of appendiceal and small bow-A uled out monthsH in advance. Because we els tumors that are incidentally found. require physician referral, this also means that some primary care providers are already V seeking out this test for their patients. Because Referrals from primary care physicians of coverageG by third-party payers, less than Although visible support from gastroen- 1%I of our patients pay out-of-pocket expens- terology and colorectal surgeryR colleagues es for the procedure. We are currently seek- can provide useful credibility for a new VC ing out additional ways to further expand program, most referrals for VC screening will our capacity, such as identifying a CT scan- actually come from primaryE care providers.Rner that will be dedicated to VC screening. Therefore, it is this group of general practitioners that must ultimately appreciate the value of VC screening. Because of theY pos- Conclusions sibility of finding notN only colorectal polyps, but also potentiallyI significant findings outside Widespread implementation of VC screen- the colon, we do not accept self-referredP ing faces multiple challenges, but these are all patients for VC evaluation. We need to ensure greatly overshadowed by the immediate need that we have the ability to directly commu- for increased participation in effective col- nicate withM a referring physician if a poten- orectal screening. Given its relatively nonin- tially significant or unusualO finding is identi- vasive nature and the wide availability of CT, fied that may need further work up. Although VC holds significant potential for addressing many primary care providers already see the a very important yet preventable public health benefit of an additionalC screening option for concern. For maximum impact, we need to their patients, others can be educated and shift our focus away from the ubiquitous updated through the use of grand rounds diminutive polyp and concentrate more on lectures and the medical literature. the larger polyps that are a more realistic

148 MINERVA CHIRURGICA Giugno 2005 VIRTUAL COLONSCOPY FOR PRIMARY SCREENING PICKHARDT

cause for concern. Although further validation tional colonoscopy for the detection of colorectal polyps. N Engl J Med 1999; 341:1540-2. of this rapidly evolving technique may be 3. Yee J, Akerkar GA, Hung RK, Steinauer-Gebauer AM, useful, clinical implementation of state-of- Wall SD, McQuaid KR. Colorectal neoplasia: perfor- the-art VC screening should proceed with- mance characteristics of CT colonography for detection in 300 patients. Radiology 2001;219:685-92. out delay by those who are willing and able 4. Johnson CD, Harmsen WS, WilsonA LA, Maccarty RL, to provide it. Welch TJ, Ilstrup DM et al. Prospective blinded evaluation of computed tomographic colonography for screen detection of colorectal polyps. Gastroenterology 2003;125:311-9. 5. Cotton PB, Durkalski VL,C Palesch YY, Palesch YY, Riassunto Mauldin PD, Hoffman B et al. Virtual colonoscopy: final results from a multi-centerI study. JAMA 2004; Colonscopia virtuale per lo screnning iniziale: il futu- 291:1713-9. ro è adesso 6. Rockey DC, Paulson EK, Niedzwiecki D, Davis W, Bosworth HB, Sanders L et al. Analysis of air contrast La colonscopia virtuale è una manovra minima- barium enema, computed tomographic colonography, mente invasiva che utilizza le moderne teconologie and colonoscopy:D prospective comparison. Lancet tomografiche computerizzate per la valutazione colon- 2005;365:305-11. 7. Pickhardt PJ, Choi JR, Hwang I, Butler JA, Puckett ML, rettale. Dalla sua introduzione, nel 1994, la colon- Hildebrandt HA et al. Computed tomographic virtual scopia virtuale ha continuato a evolversi rapidamen- colonoscopyE to screen for colorectal neoplasia in te e a migliorarsi quale strumento di screening dia- asymptomatic adults. N Engl J Med 2003;349: 2191- gnostico. Il successo iniziale ottenuto utilizzando una 200. Epub 2003 Dec 1.® rilevazione bi-dimensionale (2D) in coorti di pazien- 8. Pickhardt PJ, Taylor AJ, Johnson GL et al. Building a CT ti con molti polipi è stato disatteso nelle popolazio- colonography program: necessary ingredients for reim- bursement and clinical success. Radiology 2005;235: ni a bassa prevalenza. La successiva introduzione M17-20. della metodica tri-dimensionale (3D) endoluminale 9. Bond JH. Update on colorectal polyps: management per il rilevamento del polipo primitivo e l’utilizzo di and follow-up surveillance.T Endoscopy 2003;35: un mezzo di contrasto per via orale ha trasformato la S35-40. colonscopia virtuale in uno strumento efficace di 10. Pickhardt PJ, Choi JR. Electronic cleansing and stool tagging in CT colonography: advantages and pitfalls screening iniziale. Lo stato attuale della tecnologia encountered with primary three-dimensional evalua- relativa alla colonscopia virtuale ha già dimostratoA tion. AJR 2003;181:799-805.H di essere una via percorribile quando viene associa- 11. Hinshaw JL, Taylor AJ, Jones DA, Pickhardt PJ. ta alle attuali tecniche di colonscopia che utilizzano Prospective blinded trial comparing single and double fibre ottiche. Gli ulteriori perfezionamenti dello scree- dose sodium phosphate for virtual colonoscopy bow- V el preparation. Paper presented at the 2005 Meeting of ning con colonscopia virtuale potranno rispondere a theG Society of Gastrointestinal Radiologists, 2005, ulteriori sfide, ma al momento vi è necessità imme- February 27-March 4, San Antonio, Texas. diata di uno screening colonrettale efficace. Data la 12.I Iannaccone R, Laghi A, Catalano C, Mangiapane F, sua natura relativamente non invasivaR e l’ampia dispo- Lamazza A, Schillaci A et al. Computed tomographic nibilità della tomografia computerizzata, la colon- colonography without cathartic preparation for the detection of colorectal polyps. Gastroenterology scopia virtuale rappresenta un potenziale significati- 2004;127:1300-11. vo per rispondere a un aspetto molto importante del-R13. Pickhardt PJ. CT colonography without catharsis: the la prevenzione della salute pubblica.E Questa revisio- ultimate study or useful additional option? Gastro- ne si occuperà delle attuali tecniche di colonscopia vir- enterology 2005;128:521-2. tuale, confronterà gli studi clinici multicentrici condotti 14. Pickhardt PJ. Virtual colonoscopy to screen for colorectal su questa tecnica, discuterà gli aspetti relativi allo cancer (reply). N Engl J Med 2004;350:1148-50. Y 15. Pickhardt PJ. Three-dimensional endoluminal CT screening iniziale eseguitoN con essa e si occuperà colonography (virtual colonoscopy): comparison of brevemente degli aggiornamenti sui nostri program- three commercially available systems. AJR 2003;181: mi relativi alla colonscopiaI virtuale. 1599-606. P 16. Pickhardt PJ. Differential diagnosis of polypoid lesions Parole chiave: Colonscopia virtuale - Colon patologie seen at CT colonography (virtual colonoscopy). - Retto patologia - Tomografia computerizzata. Radiographics 2004;24:1535-59. 17. Pickhardt PJ, Nugent PA, Choi JR, Schindler WR. Flat colorectal lesions in asymptomatic adults: implications M O for screening with CT virtual colonoscopy. AJR 2004; References 183:1343-7. 18. Pickhardt PJ, Lee AD, McFarland EG, Taylor AJ. Linear 1. Vining D, Gelfand D. Noninvasive colonoscopy using polyp measurement at CT colonography: in vitro and helical CT scanning, 3D reconstruction, and virtual in vivo comparison of two-dimensional and three- reality. Paper presentedC at the 1994 Meeting of the dimensional displays. Radiology. In press. 2005. Society of Gastrointestinal Radiologists, 1994, February 19. Pickhardt PJ. Translucency rendering in 3D endolu- 13-18, Maui, Hawaii. minal CT colonography: a useful tool for increasing 2. Fenlon HM, Nunes DP, Schroy PC III, Barish MA, Clarke polyp specificity and decreasing interpretation time. PD, Ferrucci JT. A comparison of virtual and conven- AJR 2004;183:429-36.

Vol. 60, N. 3 MINERVA CHIRURGICA 149 PICKHARDT VIRTUAL COLONSCOPY FOR PRIMARY SCREENING

20. McFarland EG, Zalis ME. CT colonography: progress Lack of spontaneous regression of tubular adenomas toward colorectal evaluation without catharsis. in two years of follow-up. Am J Gastroenterol Gastroenterology 2004;127:1623-36. 1997;92:1117-20. 21. Pickhardt PJ. CT colonography (virtual colonoscopy) for 33. Loeve F, Boer R, Zauber AG. National Polyp Study data primary colorectal screening: challenges facing clinical evidence for regression of adenomas. Int J Cancer implementation. Abdom Imaging 2005;30:1-4. 2004;111:633-9. 22. Pickhardt PJ. By-patient performance characteristics 34. CT colonography reporting and dataA system (C-RADS): for CT colonography: importance of polyp size thresh- a consensus statement. (Working Group on Virtual old data. Radiology 2003;229:291-3. Colonoscopy). Radiology. In press. 2005. 23. Pickhardt PJ. Virtual colonoscopy - finally a credible 35. Pickhardt PJ, Choi JR, Nugent PA, Schindler WR. The contender (comment). Gastroenterology 2004;126: effect of diagnostic confidence on the probability of 1911-2. optical colonoscopic confirmationC for potential polyps 24. Gatto NM, Frucht H, Sundararajan V, Jacobson JS, detected at CT colonography: prospective assessment Grann VR, Neugut AI. Risk of perforation after in 1339 asymptomatic Iadults. AJR 2004;183:1661-5. colonoscopy and sigmoidoscopy: a population-based 36. Hara AK, Johnson CD, MacCarty RL, Welch TJ. study. J Natl Cancer Inst 2003;95:230-6. Incidental extracolonic findings at CT colonography. 25. Rankin GB. Indications, contraindications, and com- Radiology 2000;215:353-7. plications of colonoscopy. In: Sivak MV editor. 37. Gluecker TM, Johnson CD, Wilson LA, Maccarty RL, Gastrointestinal endoscopy. Philadelphia: WB Saunders; Welch TJ, VannessD DJ et al. Extracolonic findings at 1987. p.873-8. CT colonography: evaluation of prevalence and cost in 26. Stryker SJ, Wolff BG, Culp CE, Libbe SD, Ilstrup DM, a screening population. Gastroenterology 2003;124: MacCarty RL. Natural history of untreated colonic 911-6. polyps. Gastroenterology 1987;93:1009-13. 38. EdwardsE JT, Wood CJ, Mendelson RM, Forbes GM. 27. Bond JH. Clinical relevance of the small colorectal Extracolonic findings at virtual colonoscopy: implica- polyp. Endoscopy 2001;33:454-7. tions for screening programs.® Am J Gastroenterol 28. Pickhardt PJ, Choi JR, Hwang I, Schindler WR. 2001;96:3009-12. Nonadenomatous polyps at CT colonography: preva- 39. Hellström M, Svensson MH, Lasson A. Extracolonic lence, size distribution, and detection rates. Radiology and incidental findings at CT colonography (virtual 2004;232:784-90. Mcolonoscopy). AJR 2004;182:631-8. 29. Hofstad B, Vatn MH, Larsen S, Osnes M. Growth of 40. Pickhardt PJ, Taylor AJ. Extracolonic findings identified colorectal polyps: recovery and evaluation of unre- in asymptomatic adultsT at screening CT colonography. sected polyps of less than 10 mm, 1 year after detec- AJR. In press. 2005. tion. Scand J Gastroenterol 1994;29:640-5. 41. Pickhardt PJ, Nugent PA, Mysliwiec PA, Choi JR, 30. Hofstad B, Vatn MH, Andersen SN, Huitfeldt HS, Schindler WR. Location of adenomas missed at optical Rognum T, Larsen S et al. Growth of colorectal polyps: colonoscopy. Ann Intern Med 2004;141:352-9. redetection and evaluation of unresected polyps Afor a 42. Ramsey Sd,H Mandelson MT, Etzioni R, Harrison R. period of three years. Gut 1996;39:449-56. Cancer-attributable costs of diagnosis and care for per- 31. Hoff G, Foerster A, Vatn MH, Sauar J, Larsen S. sons with screen-detected versus symptom-detected Epidemiology of polyps in the rectum and colon: recov- colorectal cancer. Gastroenterology 2003;125:1645-50. ery and evaluation of unresected polyps 2V years after 43. Barnes E. HMO pays for screening virtual colonoscopy. detection. Scand J Gastroenterol 1986;21:853-62. AuntMinnie.comG [online]. June 4, 2004. Available at: 32. Bersentes K, Fennerty MB, SamplinerR RE, Garewal HS. Ihttp://www.auntminnie.com. E R N Y I P M O C

150 MINERVA CHIRURGICA Giugno 2005