Table of Contents

Case Presentation
Introduction
Screening Tests for Colorectal Cancer
Virtual Colonoscopy
Patients Attitudes on Virtual Colonscopy
Conclusions

VIRTUAL COLONOSCOPY

Sanjib P. Mohanty
Resident Grand Rounds
April 9, 2002

Case Presentation

Mr. C is a 52 year-old African American male with a past medical history significant for diabetes mellitus, hypertension and hyperlipidemia who presents for his annual complete physical examination. He has been doing well without any complaints. He specifically denies any melena or hematochezia and has with him two stool cards that I gave him three months prior that are heme negative upon testing. His physical exam is unremarkable and his digital rectal exam is within normal limits without any masses noted. As always, I comment on how happy I am with his state of health, but in order to complete his physical exam we must address health maintenance issues.

For the last three years I have encouraged him to undergo some sort of screening examination for colorectal cancer. He has always agreed to perform annual serial fecal-occult blood card tests. However, he has adamantly refused to undergo flexible sigmoidoscopy or colonoscopy saying, “I’m not going to have those tests because they hurt and because nobody is going to stick one of those things in my rear end! I heard in the news that there is some new ‘Cat Scan’ that can tell you if you got colon cancer. How about sending me for one of those?”

Clinical Question: In adults at average risk for colorectal carcinoma, is virtual colonoscopy a reasonable screening test for the detection of colorectal cancer?

Introduction

Colorectal cancer is the second leading cause of cancer related deaths in the United States with almost 138,200 new cases diagnosed and 55,300 related deaths each year. This accounts for approximately 9% and 11% of cancer associated mortalities in men and women, respectively. Globally, it is the third most common cancer in males and females. Approximately 6% of the American population will eventually develop colon or rectal cancer and over 6 million Americans who are alive today will die of the disease (1).

Screening and surveillance are cost effective strategies for reducing the incidence and mortality due to colorectal cancer. Yet, fewer than 30% of eligible patients have had a screening test for colorectal cancer whereas over 70% of women over the age of 40 have had a screening mammogram for breast cancer and 80% of these women have had a screening Papanicolaou smear for cervical cancer. Colorectal cancer screening in the United States is being promoted by all major medical organizations, including the American Cancer Society and the American Gastroenterological Association. These new public health policies are based on the accepted theory that most colorectal cancers arise from preexisting adenomatous polyps that remain clinically dormant until presentation, often as advanced and incurable malignant disease (2).

There are certain groups that have a high incidence of colorectal cancer. These include those with hereditary conditions, such as familial adenomatous polyposis as well as autosomal dominant nonpolyposis colorectal cancer. Together they account for approximately 6% of colorectal cancers. More common conditions associated with an increased risk include: a personal history of colorectal cancer or adenomas, first degree relative with adenomas diagnosed before 60 years of age, a personal history of ovarian, endometrial, or breast cancer, and a personal history of longstanding chronic ulcerative colitis or Crohn’s colitis. These high-risk groups account for 25% of all colorectal cancers (3).

Colorectal cancers appear to arise from adenomatous polyps which progress in size from small to large polyps (> 1.0 cm) and then to dysplasia and cancer. This usually evolves over a ten year time period in most people. A strong correlation exists between an individual’s environmental exposures and the development of colorectal cancer. Animal models have demonstrated higher rates of carcinogen-induced colonic adenocarcinomas with diets containing high amounts of polyunsaturated and saturated fat. It is hypothesized that carcinogens consumed and introduced into the bowel act with bile acid as well as other “tumor promoters” to affect epithelial cells in the colonic mucosa. It is thought that certain of these epithelial cells are genetically susceptible to carcinogenesis either by hereditary disposition or some other genotoxic events. Hence, these “immortal” cells undergo a cascade of events that forever perpetuates uncontrolled growth and the development of adenomatous polyps leading to colon cancer (4).

Screening Tests for Colorectal Cancer

The choice of which colorectal cancer screening test to use remains controversial. There are currently four Medicare-reimbursable tests for colon cancer screening: the fecal occult blood test, flexible sigmoidoscopy, the double-contrast barium enema and colonoscopy. However, each of these screening tests have various strengths and weaknesses lending to the ongoing debate about the most effective test for the detection of colorectal carcinoma. Fecal occult-blood tests are the most common screening tests for colorectal cancer in the Unites States and is the only proven screening strategy to reduce death rates by 15 to 33 percent from colorectal cancer by randomized clinical trials. A colonoscopy is usually recommended if any of three serially obtained fecal occult blood screening cards is positive as the probability of finding an adenoma or colorectal cancer is 17 to 46 percent.

Flexible sigmoidoscopy is generally performed without sedation and requires the patient to use a laxative prior to the procedure. It is highly specific and sensitive for lesions in the distal or lower half of the colon. However, sigmoidoscopy misses almost 50 percent of lesions that are in the proximal or first half of the colon. The ability for flexible sigmoidoscopy to detect proximal neoplasms relies on the presence of adenomas in the distal colon and, subsequently, the operator's decision to initiate a full colon evaluation. In these cases the sensitivity of detecting neoplasms rises to 70 percent with the use of screening sigmoidoscopy. Unfortunately, there are no randomized trials to date that have assessed sigmoidoscopy as a screening test. Certainly, sigmoidoscopy is easier to perform than colonoscopy and carries less cost and risks.

Colonoscopy is the current gold standard for colorectal cancer screening and is associated with higher costs and risks than sigmoidoscopy, but can survey the entire colon and can be used to remove adenomas or neoplasms. Medicare now reimburses for screening colonoscopies every ten years in average risk patients which is an attractive feature for this mode of surveillance. Yet, just as with flexible sigmoidoscopy there are no randomized clinical trials to support this test as the universally accepted modality for colorectal cancer screening. However, colonoscopy has been shown to be more sensitive than flexible sigmoidoscopy combined with fecal occult-blood tests making this the more "common sense" strategy.

Barium enemas have been a common alternative for colorectal cancer screening in the past but has lost some appeal as a screening test given its relative low sensitivity of 48 percent for revealing polyps larger than 1cm (2).

Virtual Colonoscopy

Computed tomographic colonography or virtual colonoscopy is a developing technology that has recently received significant attention in the national media as a possible new screening test for colorectal cancer. There have been several articles published on the subject in newspapers and magazines with an article in the October, 2001 issue of the US News and World Report proclaiming that with virtual colonoscopy “all of the risk and discomfort of a colonoscopy is eliminated, while equally important, the physician’s ability to visualize polyps is greatly enhanced.” On March 20^th and 21^st the Today Show profiled this new technology in a two part series culminating in the show’s host Katie Couric undergoing the procedure herself. With this amount of national exposure the merits of virtual colonoscopy certainly deserve further scrutiny as a potential screening test for colorectal cancer.

First described by David Vining, David Gelfand and colleagues in 1994 at Wake Forest University Baptist Medical Center, computed tomographic colonography or virtual colonoscopy is helical or spiral CT scanning of the abdomen and with application of the algorithms of virtual-reality systems to build three dimensional models of the inner surfaces of the colon. These images are then interpreted by a radiologist who survey the colon lumen in any given direction (5). Proponents of virtual colonoscopy state that the procedure is less invasive, requires no sedation, and causes less discomfort to the patient when compared to conventional colonoscopy.

Thorough bowel preparation is necessary for an accurate examination either with an oral colonic lavage (polyethylene glycol solution, magnesium citrate, sodium phosphate, etc.) the day before the procedure. On the day of the study, the patient is placed on the CT table, usually in the left lateral decubitus position. Air or carbon dioxide is introduced into the colon through a small, soft-tipped tube inserted an inch into the rectum. Smooth muscle relaxants such as glucagon is often used to reduce peristalsis. Because of the presence of stool, fluid or bowel spasm, the patient is placed in both the prone and supine positions on the CT table. These positions redistribute fluid and colonic gas to facilitate polyp detection while an uninterrupted volume of data is acquired through the abdomen during a single 32-second breath-hold. In heavier patients or patients unable to hold their breath, two scans must be performed which could potentially lead to image misregistration when the data is reformatted. Helical CT scanning parameters are set to optimize subsequent image reformation by using thin 5 mm slices to cover the abdomen and pelvis. Once image data are acquired, postprocessing is performed on a computer using virtual reality software to render multiplanar reformatted images, mucosal relief profiles, or volume-rendered endoluminal perspectives. Volumetric CT data in the entire colon are acquired with only a few seconds to minutes of scanning and with a total time of 20 minutes. However, thorough evaluation of the images after data acquisition can range in time from 15- 45 minutes depending on observer proficiency (6). The cost of performing virtual colonoscopy at Wake Forest University Baptist Medical Center is $750 (including technical fees) compared to $800-$1200 for conventional colonoscopy. Costs vary between academic centers and relative availability.

The use of virtual colonoscopy in detecting colorectal polyps and its comparison to conventional colonoscopy has been investigated in several studies. Depending on the authors and their respective background the studies and outcomes can be somewhat different. The following is a review of articles supporting virtual colonoscopy as a possible screening test for colorectal cancer.

Hara AK et. al. Detection of colorectal polyps with CT colography: initial assessment of sensitivity and specificity. Radiology. 1997

Clinical question: What is the sensitivity and specificity of CT colography (virtual colonoscopy) in detection of colorectal polyps compared to axial CT and colonoscopy?

Design: Blinded prospective study in which virtual colonoscopy followed by conventional colonoscopy was performed on 70 patients, 50 men and 20 women with age ranging from 43-80 years old. 35 patients had findings from a barium enema or flexible sigmoidoscopy that was positive for one or more polyps. 35 other patients were being followed up after removal of polyps 1-5 years earlier. This follow-up group served as the control group. Exclusion criteria included presence of a colostomy, inflammatory bowel disease, acute diverticulitis, pregnancy, claustrophobia, colon biopsy performed within 72 hours or polypectomy in the last 6 weeks.

Methods: All patients underwent standard oral colonoscopy preparation with Golytely and bisacodyl tablets before virtual colonoscopy. All virtual colonoscopy examinations were performed immediately before conventional colonoscopy with a helical CT scanner. Insufflation of the colon was performed with carbon dioxide in 60 patients while air was used in 10 patients with no difference in image quality. Glucagon was not used in any of the patients. After insufflation of the colon images were obtained with the patient in the supine position. Three to four breath holds were performed in two minutes enabling acquisition of 150 images. Two radiologists independently reviewed the axial CT images without knowledge of colonoscopy results with observers ranking their level of confidence for diagnosis on a scale of 1 to 5 (5 being most confident). To avoid learning bias, a 2 month interval was observed before the reformatted virtual colonoscopy images were reviewed again by two radiologists in a blinded fashion with no knowledge of the axial CT or colonoscopy results. All colonoscopy examinations were performed in a blinded fashion after virtual colonoscopy on the same day with 60 of the 70 cases recorded on videotape. Adenomas identified at virtual colonoscopy and not reported at colonoscopy were reevaluated by reviewing the colonoscopy videotapes. If the polyp was identified retrospectively on the videotape it was considered a colonoscopic error.

Results: A total of 161 polyps were identified at colonoscopy with 115 lesions histologically proven to be adenomas. Combining the average of the two observers’ findings on virtual colonoscopy, the sensitivity of identifying patients with adenomas 10mm or larger was 75% with a specificity of 91%. For patients with adenomas 5 mm or greater the average sensitivity and specificity were 66% and 63% respectively. However, in patients with adenomas smaller than 5 mm the average sensitivity was 42% with a specificity of 80%.

Diagnostic Performance of the Two Observers with CT Colography

Lesions > 10mm Lesions > 5 mm Lesions < 5 mm

Patients with adenomas	A	B	A	B	A	B
True Positives	9	9	17	16	13	12
False Negatives	3	3	8	9	17	18
False Positives	5	6	19	15	10	6
True Negatives	53	52	26	30	30	34
Sensitivity	75%	75%	68%	64%	43%	40%
95% CI	43-95	43-95	47-85	43-82	26-63	23-59
Specificity	91%	90%	58%	67%	75%	85%
95% CI	81-97	79-96	42-72	51-80	59-87	70-94

Conclusions: The authors suggest that virtual colonoscopy has important screening potential and performs at a sensitivity and specificity that are attractive for colorectal screening applications. They suggest that the majority of false-negative and false-positive lesions may be reduced with experience and improved scanning technique as most of these errors were secondary to poor bowel preparation, artifact or perceptual error (7).

Limitations: This is a small prospective study where all the patients that received virtual colonoscopy were at high risk (they were already found to have polyps from an earlier barium enema or flexible sigmoidoscopy, compared with patients in postpolypectomy follow-up). This study also did not assess evaluation of costs and/or patient acceptability. Furthermore, diagnostic limitations in this study included detecting adenomas less than 5 mm as well as flat adenomas.

Fenlon HM et al. A comparison of virtual and conventional colonoscopy for the detection of colorectal polyps. New England Journal of Medicine. 1999.

Clinical question: Does virtual colonoscopy detect colorectal polyps accurately in adults at high risk for colorectal neoplasia?

Design: Prospective blinded study in a Boston, Ma medical center of 100 adults (60 men and 40 women) who were 50 to 77 years of age (mean age, 62 years) and at high risk for colorectal neoplasia (> 50 years of age with a history of adenomatous polyps, recent sigmoidoscopic evidence of > 1 polyp, a positive fecal occult blood test, or a history of colorectal cancer in > 1 first-degree relatives). Exclusion criteria for participation were evidence of bowl ischemia or obstruction; retained Barium on scout film; colonic biopsy or polypectomy in the previous 14 days; allergy to glucagon; known glucagonoma, insulinoma, or pheochromocytoma; or pregnancy.

Methods: Virtual colonoscopy was done immediately prior to conventional colonoscopy after undergoing appropriate bowel preparation and receiving sodium phosphate enemas. A rectal tube was inserted, and the colon was insufflated with air to the maximal level tolerated by the patient. To minimize smooth muscle spasm and peristalsis as well as to reduce discomfort, glucagon 1 mg was given prior to thin section helical computed tomography with the patient in the prone and supine positions. Images were obtained during a single breath-holding session when possible in order to visualize the entire colon. The axial two-dimensional CT images and endoluminal three dimensional reconstructions were reviewed by two experienced gastrointestinal radiologists who were unaware of the results of conventional colonoscopy and all previous evaluations including flexible sigmoidoscopy. The radiologists reviewed the studies jointly and arrived at a consensus opinion. Conventional colonoscopy was performed immediately after virtual colonoscopy by a experienced gastroenterologist using a standard colonoscope. The gastroenterologists were unaware of the results of the virtual colonoscopy. All lesions were documented and the sizes were measured by comparison with an open biopsy forceps.

Results: 115 polyps were found in 49 patients with 101 polyps removed successfully for histologic analysis. Virtual colonoscopy detected 71% of 115 polyps and more large polyps were detected than smaller ones with sensitivities as follows: 91% for polyps > 10mm, 83% for 6-9mm polyps, and 55% for polyps < 5mm. There were three carcinomas detected by conventional colonoscopy that were also detected during virtual colonoscopy. However, there were 19 false positive findings of polyps by virtual colonoscopy ( 9 1-5mm; 8 6-9mm and 2 >10mm). Nine of the false positive findings were in segments of colon containing residual solid stool and eight were in segments of the colon where there was significant diverticular disease or complex haustral folds with poor air insufflation, hence leading to misinterpretations as polyps. Of the 33 polyps that were not detected on virtual colonoscopy, 24 were between 1-5mm in size. Limited image resolution accounted for most of the false negative results. However, false negative results for the seven polyps that were 6 to 9mm and the two that were > 10 mm were due to misinterpretation of polyps as stool, haustral folds, or retained intraluminal fluid.

Conclusions: In this study population of adults at higher risk for colorectal carcinoma virtual colonoscopy detected 91% of large (>1 cm) polyps which are the ones most likely to progress to cancer. The authors conclude that virtual colonoscopy is more accurate than barium enema (on the basis of sensitivity data of barium enema in other studies) and that it compares to conventional colonoscopy for the detection of polyps larger than 5 mm. Hence, virtual colonoscopy is a promising new choice for colorectal cancer screening and its performance will only improve with the development of CT technology (8).

Limitations: This was a small prospective study in high-risk patients which could lend to an overestimation of its performance in persons with average risk as some of these patients had recent sigmoidoscopic evidence of one or more polyps. The study design also did not allow for comparison of colonoscopy to virtual colonoscopy in terms of patients preferences, costs, side effects or tolerance of the procedure which all are important in choosing an appropriate screening test.

Yee J et. al. Colorectal neoplasia: Performance characteristics of CT colonography for detection in 300 patients. Gastrointestinal Imaging. 2001.

Clinical Question: What is the sensitivity and specificity of CT colonography for colorectal polyp and cancer detection using colonoscopy as the reference standard?

Design: Prospective blinded study of 300 patients (291 men, 9 women; age range 25-90 years with mean age of 62.6 years) referred for standard colonoscopy for colorectal cancer screening or for evaluation of symptoms including iron deficiency anemia, heme-positive stools, hematochezia or personal or family history of colon cancer. Pregnancy was the only exclusion criteria. CT colonography was performed 2-3 hours prior to conventional colonoscopy.

Methods: CT colonography was performed 2-3 hours prior to conventional colonoscopy after maintaining a clear liquid diet and drinking 10 oz. of magnesium citrate and 4 L of polyethylene glycol solution (Colyte) The patient was placed supine on the CT table and a rectal tube was inserted and the colon was insufflated with air to the maximal level tolerated by the patient (average of 30-40 bulb compressions). A subset of 115 patients received 1.0 mg of glucagons to minimize smooth muscle spasm and peristalsis as well as to reduce discomfort. Thin section helical CT images were obtained during a single breath-hold from the top of the colon through the rectum. The patient was then placed in the prone position and repeat scanning was performed with the same protocol. The 2-D CT reformatted images and endoluminal reconstructions were reviewed by two radiologists independently and a consensus reading was performed. The radiologists were unaware of the patients’ clinical history or results of conventional colonoscopy. Conventional colonoscopy was performed immediately after CT colonography by an experienced gastroenterologist using a standard colonoscope. The gastroenterologists were unaware of the results of the CT colonography. All lesions were documented and the sizes were measured by comparison with an open biopsy forceps. The findings at CT colonography were compared by using two different methods: direct by-polyp comparison (polyp noted in one study had to be in the same or adjacent segment with less than 4 mm difference in size) and by-patient comparison (polyps at CT colonography were considered to match if both studies demonstrated at least one polyp or neither showed a polyp regardless of location or size).

Results: In 182 patients colonoscopy detected 524 polyps and eight infiltrative or annular masses. Of these 524 polyps, 289 were adenomatous during histological examination. Similarly, the 8 annular masses were confirmed to be carcinoma with histology. CT colonography performance was as follows:

CT colonography had 100% sensitivity (8/8) for the detection of carcinomas.
Using the direct by-polyp matching criteria the sensitivity for overall polyp detection was 69.7% (365/524). The sensitivity was 90% (74/82) for the detection of polyps 10 mm or larger, 80.1% (113/141) for polyps 5.0-9.9 mm and 59.1% (178/301) for polyps smaller than 5 mm. CT colonography demonstrated 185 false-positive polyps in 113 patients. Most of these (142) occurred in poorly distended or poorly prepared segments of the colon.
The sensitivity for adenoma detection for polyps 10mm or larger was 94% (64/68); 82% (72/88) 5.0-9.9 mm; and 66.9% (95/142) for adenomas smaller than 5 mm.
When using the by-patient comparison the overall sensitivity for polyp detection was 90.1% (164/182) and for adenomas was 93.9% (124/132). The specificity for polyp detection was 72.0% (85/118).

Sensitivity of CT Colonography for Lesion Detection with the Bi-Polyp Comparison

Lesion size (mm)	Sensitivity for polyps (%)	Sensitivity for adenomas (%)
Overall	69.7 (365/524)	77.5 (231/298)
< 5.0	59.1 (178/301)	66.9 (95/142)
5.0-9.9	80.1 (113/141)	82 (72/88)
>10.0	90 (74/82)	94 (64/68)
Cancer	100 (8/8)	NA

Sensitivity and Specificity of CT Colonography for Lesion Detection with the By-Patient Comparison

Lesion size (mm)	Sensitivity for polyps (%)	Sensitivity for adenomas (%)	Specificity for polyps (%)
Overall	90.1 (164/182)	93.9 (124/132)	72.0 (85/118)
< 5.0	82 (65/79)	86 (37/43)	8 false positives
5.0-9.9	93 (50/54)	95 (40/42)	23 false positive
>10.0	100 (49/49)	100 (47/47)	2 false positives
Cancer	100 (8/8)	NA	NA

Conclusions: CT colonography is comparable with standard colonoscopy for the detection of clinically important polyps and cancer. The authors state that since polyp size is directly related to the probability of malignancy, a sensitivity of 94% for clinically important polyps (adenomas 10 mm or larger) means that only 6% of clinically important polyps were missed in this study (9).

Limitations: The study population consisted of mostly men which does not represent the general screening population. The study also failed to expand on their direct by-polyp results which had significantly lower sensitivities per polyp size. The study design also did not allow for comparison of colonoscopy to virtual colonoscopy in terms of patients preferences, costs, side effects or tolerance of the procedure which all are important in choosing an appropriate screening test.

The following are several studies that were designed in much the same fashion but the results were not as supportive of virtual colonoscopy as a screening test for colorectal cancer. The studies all appeared in gastroenterology journals and the primary authors were all gastroenterologists.

Rex, DK et. al. An initial experience with screening for colon polyps using spiral CT with and without virtual colonoscopy. Gastrointestinal Endoscopy. 1999.

Clinical question: What is the sensitivity of virtual colonoscopy for polyp detection in an asymptomatic population including persons without colon neoplasia?

Design: Blinded, prospective study in which 46 patients (44 men) older than 60 years without symptoms and no history of colon polyps were recruited from the Richard A. Roudebush Veterans Affair Hospital in Indianapolis. 29 of these patients had never undergone a colon examination while 17 had prior screening flexible sigmoidoscopies that revealed polyps. These demographic features were selected to increase the prevalence of polyps.

Methods: All subjects were prepared with standard colonoscopy preparations (Fleet Phospha-Soda or Golytely) the day before the study. All patients were placed on the CT table in the left lateral decubitus position and a catheter was placed in the rectum. Glucagon 1 mg was given intravenously and carbon dioxide was introduced into the rectum until the patient had a feeling of discomfort or abdominal fullness (mean volume was 2 liters). Spiral CT was performed using a conventional dual slice technique. The patient was then turned into the prone position and the spiral CT imaging was repeated. CT data sets were processed using Bowman Gray virtual endoscopy software generating volume-rendered images of the colon. Colonoscopy was performed after the spiral CT with removal of all polyps. The two and three dimensional images were then read in batches of 27 cases, 16 cases, and 3 cases. After each batch was read the colonoscopic findings were revealed to the radiologists.

Results: 33 of the 46 patients had adenomas as documented by histology. This included 14 of 17 patients who had known polyps on prior flexible sigmoidoscopy. A total of 91 polyps were detected by colonoscopy. Conventional spiral CT detected 10 of these polyps whereas virtual colonoscopy detected 20. Interpretation of the virtual images was equal to or superior to the regular spiral CT images for all size ranges and was statistically significant for polyps in the 6-9mm range. Three of the four polyps 20 mm in diameter that were missed by virtual colonoscopy were flat adenomas of the right colon. One of these had retained stool and motion artifact in the area of the polyp. In the other two cases, the miss was in failure to perceive the polyp. Collapse of colonic segments or retained water and stool was the culprit for most of the other missed polyps. There were three false positive findings of polyps by virtual colonoscopy. Specificity for patients with adenomas 1 cm or larger was 89%.

Sensitivity of CT colography for colorectal adenomas by size

Size Range	Number of adenomas	Axial CT sensitivity	Three-dimensional sensitivity	p-Value
>20 mm	4	1 (25%)	1 (25%)	NS
10-19 mm	10	3 (30%)	6 (60%)	NS
6-9 mm	14	1 (7%)	6 (43%)	0.03
<5 mm	63	5 (8%)	7 (11%)	NS

Conclusions: Virtual colonoscopy failed to detect a significant number of large benign colon polyps and most small adenomas, especially flat adenomas of the right colon. Some of these were due to incomplete preparation of the bowel, artifact, or perceptual error. The authors do not recommend the use of virtual colonoscopy for colorectal cancer screening (10).

Limitations: This is a small prospective study of 46 patients 44 men of whom were men. The study increased the prevalence of neoplasia by selecting patients with abnormal findings at flexible sigmoidoscopy as well as selecting older aged males leading to higher incidence of adenomas.

Kay CL et. al. Virtual endoscopy- comparison with colonoscopy in the detection of space-occupying lesions of the colon. Endoscopy. (2000).

Clinical question: How dose virtual colonoscopy compare to conventional colonoscopy in patients with known or suspected space-occupying lesions?

Design: 38 patients (15 men, 23 women; age range 31-89) with a history of heme-positive stool, polyps on previous colonoscopy, prior history of colon cancer, unexplained iron-deficiency anemia or a mass on barium enema or other imaging study were scheduled for a CT scan followed by colonoscopy on the same day. Exclusion criteria included pregnancy, radiologist’s awareness of diagnosis, or patient’s refusal to participate in the study.

Methods: All patients prepared for the colonoscopy the night before with standard colonoscopy preparation with either colon lavage solution or Fleet Phosphosoda. Prior to CT scanning patients received glucagon 1mg IV to reduce bowel spasm during air insufflation and to reduce motion artifact on CT scan. Patients were placed in the left lateral decubitus position and received air insufflation as tolerated via a rectal tube. After a CT scout film was performed the patient was placed in the prone position and more air was insufflated if necessary. After appropriate colonic distention, a single helical CT scan of the abdomen and pelvis was performed in a craniocaudal direction. Three dimensional reconstructions were facilitated with the application of virtual endoscopy perspective software. Interpretation of the 3D images was performed blinded to the results of subsequent colonoscopy. The endoscopist was also unaware of the virtual colonoscopy results.

Results: Results were reported on a per-polyp basis as well as per-patient sensitivity and specificity for polyps 5-9 mm and > 10 mm in size. A total of 13 polyps 5-9 mm and 11 polyps > 10 mm were identified by conventional colonoscopy. Virtual colonoscopy detected 5 of 13 polyps in the 5-9 mm size and 10 of 11 polyps greater than 10 mm. Missed lesions on virtual colonoscopy were attributed to a collapsed rectum for 2 polyps and retained stool for the other two. Six lesions were noted on virtual colonoscopy but were not seen on conventional colonoscopy and were assumed to be false positive results.

Per-polyp sensitivity of virtual colonoscopy

	Polyps 5-9 mm	Polyps >10 mm
Conventional Colonoscopy	13	11
Virtual Colonoscopy (VC)	5	10
Per-polyp sensitivity of VC	38.5%	90.9%

Per-patient sensitivity and specificity of virtual colonoscopy

	Polyps > 5 mm	Polyps >10 mm
Sensitivity	66.7%	90.0%
Specificity	75.0%	82.1%

Conclusions: In this study virtual colonoscopy was able to detect most lesions greater than 10 mm in diameter. As the technique continues to develop, the authors believe that virtual colonoscopy should continue to be evaluated as a diagnostic and screening tool in colorectal neoplasia (11).

Limitations: The study was too small to draw significant conclusions and the patients were selected based on known colonic pathology. Furthermore, in this study only one prone scan was used versus two scans in the prone and supine positions as is the case in most other studies.

Spinzi G et. al. Computed tomographic colonography and conventional colonoscopy for colon diseases: a prospective, blinded study. Am Journ Gastro. 2001.

Clinical question: What is the sensitivity, specificity and diagnostic accuracy of virtual colonoscopy compared with colonoscopy?

Design: Prospective blinded study in one single institution in Italy of 99 patients randomly selected among those already scheduled to receive a diagnostic colonoscopy. Exclusion criteria were as follows: <18 years old, inability to give written consent, clinical indication for partial colonoscopy; refusal to participate; clinical suspicion of acute diverticulitis; serious claustrophobia, or serious concomitant disease.

Methods: CT colonography was offered to the first patient attending the endoscopic unit each morning. If that patient refused or was found ineligible, the next patient was offered the procedure. The first 49 patients underwent colonoscopy before virtual colonoscopy. The radiologist did not know the findings of the colonoscopy. In the second phase, 50 patients were examined by virtual colonoscopy first followed by colonoscopy with the gastroenterologist unaware of the virtual colonoscopy results. Polyps and neoformations found by virtual colonoscopy were compared to those found on colonoscopy by a supervisor who reviewed written reports when all the data had been collected.

Results: Of the 99 patients enrolled for virtual colonoscopy, 3 patients were unable to complete the study because of incomplete colonoscopy (extrinsic mass in one patient and persistent looping of the scope in another) or because of incomplete CT colonography because of improper colonic distention. Hence, 96 patients were fully assessed by both methods. Colonoscopy diagnosed eight tumors (confirmed histologically) and virtual colonoscopy found seven missing one tumor in the rectum in a patient whose bowel preparation was inadequate. Colonoscopy identified 45 polyps in 28 patients. Only 26 polyps were detected by virtual colonoscopy with a sensitivity of 57.8% and 96.2% specificity. 13 polyps (9 polyps <10 mm and 4 >10 mm) were not identified because of inadequate bowel preparation and insufficient distention. 6 polyps were missed because of errors committed when reading the virtual colonoscopy scans. Virtual colonoscopy also gave four false-positive findings among polyps < 10 mm. Virtual colonoscopy also failed to detect one 8 mm flat adenoma in the cecum, telangiectasisas and colonic lesions in a patient with portal hypertension; it also missed ulcers in the left colon in two patient’s with Crohn’s disease and signs of ulcerative proctosigmoiditis in three patients.

Conclusions: This study is the only study to be conducted in a blind fashion on a random unselected population. Overall, virtual colonoscopy diagnosed only 57.8% of the polyps, regardless of size. Inadequate bowel preparation appeared to be a major factor in false negative results. Virtual colonoscopy also failed to detect a flat adenoma as well as other flat lesions such as telangiectasias and inflammatory bowel disease lesions. Hence, the authors state that although virtual colonoscopy offers the potential advantage of being safer and less invasive, it lacks the necessary sensitivity and specificity for routine colorectal cancer screening (12).

Limitations: This was another small study population. The sensitivity in identification of polyps would have been improved if the study had excluded from analysis the four patients with collapse of colonic segments or with stool in the areas that harbored 13 missing polyps. Also, the radiologists’ proficiency in reading the scans improved as the study moved forward as the last 20 examinations demonstrated a sensitivity of 91.6% compared to 31.8% in the first 20 studies.

Learning Curve for Virtual Colonoscopy

Patients Polyps found by colonoscopy Polyps found by virtual colonoscopy (sensitivity)

1-25	22	7 (31.8%)
26-50	8	6 (75%)
51-75	3	2 (66.6%)
76-96	12	11 (91.6%)

Patients Attitudes on Virtual Colonoscopy

There is significant disparity in patient attitudes toward colorectal cancer screening. Patient reticence to undergo screening may be due to the fear of discomfort, the diagnosis of cancer or the belief that the detection of cancer will not change their lifespan. Poor patient acceptance of colonoscopic screening has also been attributed to the colon cleansing required for colonoscopy. Proponents of virtual colonoscopy hope that some of these issues would be resolved. Virtual colonoscopy may be safer and could be less uncomfortable because it is noninvasive and does not require sedation. The following studies addressed some of these issues.

Akerkar GA et al. Patient experience and preferences toward colon cancer screening: a comparison of virtual colonoscopy and conventional colonoscopy. Gastrointestinal Endoscopy. 2001

Clinical question: What is the difference in patient’s’ preferences and tolerances of colonoscopy compared to virtual colonoscopy?

Methods: 295 patients (287 men and 8 women; mean age 62.4 years) at a veterans affairs hospital were scheduled to undergo virtual colonoscopy followed by conventional colonoscopy for either evaluation of clinical symptoms (191 patients) or for screening (104 patients). They were all were given a questionnaire that they were to complete after virtual colonoscopy, after colonoscopy and again 24 hours after completing colonoscopy (to assess the validity of their initial responses). The questionnaire addressed overall pain, sense of respect, and discomfort using a 7-point Likert scale, with the higher numbers denoting a worse experience. Patients were also asked to choose the test they most preferred using a time tradeoff technique. This techniques assesses the length of time (one, two, three, four, >8, or >26 weeks) a patient would be willing to wait for the test of their choice if they could undergo the other test immediately.

Results: All 295 patients completed the questionnaire immediately after the procedures and only 83 patients completed the questionnaire at 24 hours. The results from the two questionnaires are as follows:

1) 295 patients reported a mean pain score of 3.61during virtual colonoscopy (VC) compared with a mean score of 1.57 during conventional colonoscopy (CC) (p<0.01). The mean discomfort score after VC was 3.19 compared with 1.45 after CC (p<.01). Patients also felt less respected with a mean score of 1.78 after VC compared to 1.18 after CC (p<.01). Additionally, patients preferred CC to VC and quantified this preference by their willingness to wait an average of 4.9 weeks for conventional colonoscopy.

2) 83 patients reported a mean pain score of 2.58 after VC compared to 1.92 after CC (p<0.01); mean discomfort score was 2.10 after VC versus 1.80 after CC (p<0.01). No significant difference was found in the level of respect score after VC, 1.56 compared to 1.49 for CC (p>.05). However, in the questionnaire obtained 24 hours after the procedures, patients reported that would wait an average of one week to undergo virtual colonoscopy rather than conventional colonoscopy.

Mean Scores After Procedures

Procedure	Pain	Discomfort	Lack of Respect
Virtual Colonoscopy	3.61	3.19	1.78
Colonoscopy	1.57	1.45	1.18

Mean Scores After 24 Hours

Procedure	Pain	discomfort	Lack of respect
Virtual Colonoscopy	2.58	2.10	1.56
Colonoscopy	1.92	1.80	1.49

Conclusions: The authors suggest that virtual colonoscopy is an attractive new method for screening for colorectal neoplasia, but improvements in patient tolerance and acceptance will be necessary for this test to become a plausible alternative for colon cancer screening (13).

Limitations: This study is purely based on the subjective dimensions of patient experience which are inherently difficult to measure. Since all the patients received sedation for the colonoscopy, the post-procedure responses could be greatly altered. Additionally, this study was composed of predominantly men at a veterans affairs hospital which does not represent the general screening population. Another limitation was that only 83 of 295 patients responded at 24 hours which may not adequately represent the true study population. In a response letter to this article submitted to GI Endoscopy Dr. B. Pineau and Dr. D. Vining of Wake Forest University Baptist Medical Center write, “considering that midazolam has a half-life of 1.3-2.2 hours, the information captured by these questionnaires was most likely obtained at a time when patients were chemically impaired… a 28% response rate (for the 24 hour mail-in response) is clearly inadequate and prevents any conclusions to be drawn from this data” (14).

Angtuaco TL et. al. Differing attitudes toward virtual and conventional colonoscopy for colorectal cancer screening: surveys among primary care physicians and potential patients. American Journal of Gastroenterology (2000)

Clinical Question: What are the prevailing attitudes among primary care physicians and patients concerning virtual colonoscopy compared to conventional colonoscopy for colorectal cancer screening?

Methods: One thousand randomly selected primary care physicians across the United States were sent questionnaires either by e-mail or regular mail comparing various aspects of virtual colonoscopy and conventional colonoscopy. Four hundred consecutive customers at a local Chicago video store were personally administered a similar questionnaire. This site was specifically chosen because it represented the demographics of the general screening population in the United States. The questionnaires contained progressively detailed information comparing the two screening tests and asked for the respondents choices based on the information presented. Data was analyzed and comparisons made based on the respondents preferences.

Results: 188 of 1000 primary care physicians and 323 of 400 completed and returned the questionnaire. Considering all the information on the survey 60.2% of potential patients and 30.3% of physicians preferred virtual colonosocpy. 25.7% of potential patients and 44.9% of physicians preferred colonoscopy. 23.6% of potential patients and 52.9% of physicians valued the ability of colonoscopy to visualize the mucosa directly. 67.4% of potential patients and 51.6% of physicians preferred virtual colonoscopy because of the lack of sedation. Finally, 82.3% of potential patients would participate in colorectal cancer screening and 61.7% of physicians would refer more for colorectal cancer screening if virtual colonoscopy was available.

Conclusions: Potential patients preferred virtual colonoscopy to conventional colonoscopy while primary care physicians preferred conventional colonoscopy for its ability to visualize the mucosa as well as its therapeutic ability (15).

Limitations: The response rates for primary care physicians were low compared to potential patients, hence leading to nonresponse bias in that the physicians who answered the surveys are more interested and possibly possess more knowledge about the topic. Furthermore, the potential patient survey conducted at the local video store may not reflect the demographics of the general U.S. population.

Conclusions

Virtual colonoscopy is a novel, exciting, and sophisticated technique for imaging the colon and rectum. Preliminary research results have been encouraging and will surely improve with advanced technology and experience of the operators. When compared to conventional colonoscopy, virtual colonoscopy does not require sedation and is not accompanied with any risk of bowel perforation. Furthermore, it seems that more patients are receptive to the idea of undergoing virtual colonoscopy when given a choice between virtual and conventional colonoscopy. The notion of evaluating both the inner colonic lumen as well as extra-intestinal pathology makes virtual colonoscopy even more promising as a potential screening test for colorectal cancer. However, despite its promise there continue to be some issues that currently affects its use as a general screening test.

Advances need to be made in improving CT data acquisition which would provide fewer image artifacts, better resolution and shorter scanning time. The advent of scanners with greater resolution may increase accuracy in detecting small, flat adenomas, which have a high risk of being malignant when located in the right colon. Furthermore, with shorter scanning times more procedures could be performed making it a more appealing choice for a screening test.

The concept of virtual colonoscopy is initially appealing to patients, but issues of bowel preparation as well as room air or carbon dioxide insufflation of the colon without sedation can make it as uncomfortable as undergoing conventional colonoscopy. The possibility of “fecal tagging” with oral contrast agents may eventually obviate the need for bowel cleansing and could potentially reduce the false positive findings on virtual colonoscopy. However, colonic insufflation is imperative in obtaining good results.

Cost-analysis assessments of virtual colonoscopy for colorectal cancer screening will need to be investigated further. Decision processes regarding the potential use of a screening test are complex and involve estimates of performance characteristics, cost, patient acceptability, compliance, availability, and long-term outcomes. Increased compliance would decrease overall cost of colorectal cancer screening programs. To be competitive, virtual colonoscopy would need to be at the same or slightly higher cost of barium enema and at a lower cost than conventional colonoscopy. The challenge remains to implement virtual colonoscopy clinical trials in a purely screening population.

If virtual colonoscopy measures to be an accurate, reliable and cost-effective method for detecting adenomas and early carcinomas, it could potentially affect the general population’s participation in screening programs and could eventually be a useful adjunct as a first-line screening technique for colorectal cancer.

References

Burt RW. Colon cancer screening. Gastroenterology. 2000; 119: 837-853.
Ransohoff DF, Sandler RS. Screening for colorectal cancer. New England Journal of Medicine. 2002; 346(1): 40-44.
Lawrence SP, Ahnen DJ. Risk factors for the development of colorectal cancer. Up to Date. Version 9.2. 2001.
Sleisenger MH, Fordtran JS. Gastrointestinal Disease: Pathophysiology/Diagnosis/Management. Fifth Edition. WB Saunders Company. 1993.
Vining DJ, Gelfand DW. Noninvasive colonoscopy using helical CT scanning, 3D reconstruction, and virtual reality; Presented at the 23^rd Annual Meeting and Postgraduate Course of the Society of Gastrointestinal Radiologists, Maui, Hawaii, 1994.
Johnson CD, Dachman AH. CT colonography: the next colon screening examination? Radiology. 2000; 216:331-341.
Hara AK, Johnson CD, Reed JE, Ahlquist DA, Nelson H, MacCarty RL, Harmsen WS, Ilstrup DM. Detection of colorectal polyps with CT colography: initial assessment of sensitivity and specificity. Radiology. 1997; 205:59-65.
Fenlon HM, Nunes DP, Schroy III PC, Barish MA, Clarke PD, Ferrucci JT. A comparison of virtual and conventional colonoscopy for the detection of colorectal polyps. New England Journal of Medicine. 1999; 341(20):1496-503.
Yee J, Akerkar GA, Hung R, Steinauer-Gebauer AM, Wall SD, McQuaid KR. Colorectal neoplasia: performance characteristics of CT colonography for detection in 300 patients. Gastrointestinal Imaging. 2001; 219: 685-692.
Rex, DK, Vining D, Kopecky KK. An initial experience with screening for colon polyps using spiral CT with and without CT colography (virtual colonoscopy). Gastrointestinal Endoscopy. 1999; 50(3): 309-13.
Kay CL, Kulling D, Hawes RH, Young JWR, Cotton PB. Virtual endoscopy- comparison with colonoscopy in the detection of space-occupying lesions of the colon. Endoscopy. 2000; 32(3): 226-232.
Spinzi G, Belloni G, Martegani A, Sangiovanni A, Favero CD, Minoli G. Computed tomographic colonography and conventional colonoscopy for colon diseases: a prospective, blinded study. American Journal of Gastroenterology. 2001; 96(2): 394-400.
Akerkar GA, Yee J, Hung R, McQuaid K. Patient experience and preferences toward colon cancer screening: a comparison of virtual colonoscopy and conventional colonoscopy. Gastrointestinal Endoscopy. 2001; 54(3): 310-5.
Pineau BC, Vining DJ. Better tolerance for colonoscopy or intolerance to virtual colonoscopy? Submitted letter to Gastrointestinal Endoscopy.
Angtuaco TL, Banaad-Omiotek GD, Howden CW. Differing attitudes toward virtual and conventional colonoscopy for colorectal cancer screening: surveys among primary care physicians and potential patients. American Journal of Gastroenterology. 2000; 96: 887-893.