Finding Polyps at Colonoscopy Previously Noted on CT Colonography

Colon screening examinations have been shown to discover neoplastic lesions at an early stage. Even the most careful studies by colonoscopy and by computed tomographic colonography (CTC) can overlook tumors with a diameter greater than 5 mm. Advances in technology have continually improved the ability to find polyps, which will lead to a real decrease in colorectal cancer incidence.

From the standpoint of the colonoscopist, it is almost unbelievable that a polyp could be reported as a positive finding on CTC but not seen on a follow-up colonoscopy. Colonoscopy is the procedure that, because of its ability to visualize the mucosal surface of the colon and delineate stool from polyps, led to the barium enema losing its status as the primary imaging tool for the large bowel. Colonoscopy has the ability to suction pools of fluid from the large bowel, to see the surface in full color, to wash away debris such as fecal material or seeds, identify any protrusion as mucosal in origin, and, with a high degree of certainty, distinguish benign from malignant lesions. Because of the visual clarity of colonoscopy, it has been hailed as the standard for colon imaging since it was first introduced.

Vast strides in radiographic imaging have seen computed tomographic colonography (CTC) become a useful tool for screening of the large bowel. A lesion that is reported on CTC but not seen on colonoscopy is often counted as a false-positive radiographic reading. CTC lesions that are seen on colonoscopy are accepted as positive findings. In general, gastroenterologists, radiologists, and surgeons view optical colonoscopy as the “gold standard” for visualizing lesions in the large bowel. Because the colonoscope is an extension of the endoscopist’s eye, having flexibility, tip deflection capability, wide-angle lenses, and the ability to detect colon pathology, it has widely replaced the barium enema (both single and double contrast) as the imaging modality for evaluation of the colon. It is not surprising that comparative studies have been performed that show the superiority of colonoscopy over the barium radiographic technique for large bowel screening. Two studies have demonstrated extremely low sensitivity and specificity for the barium enema compared with colonoscopy. The sensitivity of the barium radiographic colonography for polyp detection of any size was found to be poor, and even for polyps greater than 10 mm, the ability to find polyps on barium enema is 50% that of colonoscopy. A study that retrospectively reviewed the reports of double contrast barium enema (DCBE) within 36 months prior to the diagnosis of colorectal cancer found that the overall rate of new or missed cancers was about 22%. In this report, DCBE missed one-fifth of all subsequently surgically resected neoplasms. Several factors accounted for this apparent high rate of “missed lesions” in the colon. The 6 factors that were associated with missing the cancer were older age, female sex, previous abdominal pelvic surgery, diverticular disease, right-sided colorectal cancer, and having the radiographic examination performed in an office setting. The investigators’ conclusion was that physicians who use DCBE to evaluate the colon must inform their patients that if a cancer is present, there is an approximately 1-in-5 chance that it will be missed.

Among the many benefits of computer technology has been the emergence of new, better, faster, and more specific radiographic imaging procedures such as the CTC. It was inevitable that the two radiographic imaging studies be compared. In a recent retrospective report reviewing findings of patients with colon cancer who had either a DCBE or CTC, only 21 of 33 patients had their malignant neoplasm detected using DCBE whereas 32 of a similar cohort of 33 patients had the tumor detected on CTC.

A meta-analysis reviewed the performance of DCBE versus CTC for the detection of colon polyps greater than or equal to 6 mm using colonoscopy as a gold standard. This comparison revealed that CTC markedly increased the ability to detect 6 mm polyps, and was also more sensitive than DCBE in detecting polyps of 6 to 9 mm. The conclusion was that DCBE has statistically lower sensitivity and specificity than CTC for detecting colorectal polyps greater than or equal to 6 mm.

In an editorial-type discussion, Stevenson stated that, in comparison to DCBE radiographic examination of the colon, CTC is more accurate, is preferred by patients, has a shorter room time, fewer complications, and lower radiation exposure, and in addition reveals therapeutically significant extracolonic lesions in 5% to 10% of cases. He states that it is “rather irresponsible to continue to offer routine DCBE examinations.”

A review of the recent medical literature on comparing CTC with colonoscopy has found many published studies that purportedly evaluate head-to-head comparisons of end results in screening average or high-risk populations for the presence of polyps or carcinoma. In several of these reports, a colonoscopic examination was performed after a full CTC examination with the colonoscopist “blinded” to the results of the CTC. Because the contrast used for CTC has been found not to interfere with the colonoscopy procedure, such a blind comparison would seem to be the ideal method to identify whether the CTC has missed any lesions and, conversely, should also reveal whether lesions seen on CTC could have been missed by the subsequent colonoscopic examination. Most of these reports have adopted colonoscopy as the gold standard for evaluation of CTC findings. Only a few articles have actually examined the possibility that colonoscopy may not find a true lesion reported on the CTC examination. A meta-analysis reported on 47 articles in which CTC was compared with colonoscopy, with all using colonoscopy as a gold standard to affirm or rule out the presence of a lesion found on CTC. Several comparative reports have stated that they used the technique of “blind colonoscopy” performed after the CTC examination and that after each segment was examined by the colonoscopist, the finding on the CTC was revealed. It is unfortunate that most centers that used a “blind and revealed” protocol whereby CTC findings were given to the colonoscopist after viewing each segment have not reported on the actual number of lesions that were reported on CTC but missed on the first blind colonoscopic examination, but then found on a second pass after the finding on CTC examination was revealed.

A study group of 15 clinical sites participated in the National CT Colonography Trial of the American College of Radiology Imaging Network, designed to compare the finding of CTC with same-day colonoscopy for evaluation of large colorectal adenomas and cancers (equal to or greater than 10 mm in diameter). The colonoscopist was blinded to the CTC results but if the radiographic examination revealed a lesion equal to or greater than 10 mm in diameter and was not seen on the initial colonoscopy, the patient was advised to undergo an additional colonoscopic examination within 90 days. For the repeat colonoscopic examination, the CTC results were provided to the examiner before the examination. In a group of 2531 patients, 30 lesions measuring 10 mm or more were reported on the CTC in 27 participants, but not detected on the initial colonoscopic examination. Fifteen of these 27 patients, having 18 reported lesions, did have a second colonoscopic examination, and in this group, 5 of the 18 lesions were confirmed as true-positive CTC on the second colonoscopy examination. The diameters of these 5 lesions found on the second-look colonoscopy were 9 mm and 14 mm (inflammatory polyps), 10 mm and 11 mm (tubular adenomas), and a 35 mm tubulovillous adenoma. Overall, there were a total of 109 neoplasms (cancers or adenoma) equal to or greater than 10 mm found by CTC and eventually determined to be positive findings by colonoscopy. One of the problems of this study is that the original colonoscopic examiner was not provided with immediate feedback on the CTC location of polyps so that the area could be reexamined immediately during that procedure.

The most effective and accurate method to ensure that the CTC finding is a true positive has been addressed by Pickhardt and colleagues, who enrolled 1253 asymptomatic adults to perform same-day CTC and colonoscopy. After interpretation of the CTC examination was available and reported, a colonoscopic examination was performed on the same day in all patients. After the colonoscopist examined each segment of the colon, the CTC results were revealed. If a reported polyp was not seen during the colonoscopy, the examiner reintubated that segment of the colon with the intention to verify or completely exclude the presence of a polyp. Because of CTC limitations, there was no attempt to include any polyp that measured 5 mm or less on either CTC or colonoscopy. In this study a total of 1310 polyps were found at CTC, with 511 of these polyps measuring 5 mm or greater in diameter. Of these 511 polyps, 55 (10.8%) were found only on the second-look colonoscopy after segmental unblinding of the written CTC report. Twenty-one of these polyps were adenomas (6 mm or larger) removed from 20 patients (mean diameter of 8.1 mm; range, 6–17 mm). The adenoma miss rate on the initial blinded prospective colonoscopy examination was 10.0% (21 of 210 adenomas), measuring at/or larger than 6 mm. These 20 patients who had missed adenomas (that measured 6 mm or larger) represented 11.9% of all patients with adenomas of that size that were found and removed during colonoscopy. The histology of these “missed” neoplasms found on the second-look colonoscopy after segmental unblinding showed that 17 were tubular adenomas, 3 were tubulovillous adenomas, and 1 was a small adenocarcinoma. Fifteen of these neoplasms were sessile, 4 pedunculated, and 2 flat. Ten of the 21 missed neoplasms were located in the proximal colon and 6 of the 11 distal lesions were in the rectum. During a repeat study of the CTC examinations, the majority of the nonrectal neoplasms (14 of 15) were located on a fold with 10 on the proximal aspect or the edge of folds. One adenoma (above the rectum) that was associated with a fold was located on the inner aspect of an acute bend in the colon. Five of the 6 missed adenomas in the rectum were within 10 cm of the anal verge on CTC. This study involved 3 medical centers with 17 experienced colonoscopists (3 of the 21 findings of adenomas missed on the initial post-CTC colonoscopy were performed by colorectal surgeons and the rest by gastroenterologists). The lesson from this study is that colonoscopy can overlook polyps in the colon and that some reported lesions on CTC that are categorized as false positives by subsequent negative colonoscopy may actually exist but were overlooked on the colonoscopic examination. This report, by revealing that significant lesions may be overlooked on colonoscopy, is an important message for colonoscopists and indicates the need for continued improvements in colonoscopic technology. Areas that could be potentially “blind” to the colonoscopist are those that are on the proximal side of folds, the inner aspect of flexures, and in the rectum. In the Pickhardt series, polyps located on the proximal side of a colonic fold accounted for two-thirds of missed adenomas (above the rectum).

The Pickhardt study should not come as a surprise, because missed lesions have been reported by colonoscopists for the last several years. The first report of back-to-back colonoscopies on the same day and immediately following each other was in 1991. The next report of tandem colonoscopy appeared 6 years later, and the most recent was in 2008. The overall miss rates for adenomas in the earlier studies were 15% to 24%. The large multicenter European study found that the miss rate for all polyps was 28%, for hyperplastic polyps 31%, and for adenomas 21%. However, for those equal to or larger than 5 mm, the miss rate for all polyps was 12% and for adenomas 9%. Among the 14 polyps and 6 adenomas larger than 5 mm missed during the first examination, 5 polyps and 1 adenoma were sessile, 9 polyps and 5 adenomas were flat. Thirty-seven adenomas were overlooked in 286 patients with the median size being 3 mm; however, the range was from 1 to 18 mm. Three advanced adenomas were missed with a size from 15 to 18 mm. In this study, which reported a 27% rate of missed adenomas for lesions less than 5 mm in diameter, the miss rate for lesions greater than 5 mm in diameter was 9%. In a previous study of 183 patients having tandem colonoscopy, Rex and colleagues reported a 27% miss rate for polyps smaller than 6 mm in diameter and only 6% for polyps larger than 9 mm. This rate represented 2 patients whose polyps were detected on a repeat colonoscopic examination. Benson and colleagues evaluated the polyp miss rate on repeat colonoscopic examinations at 4 months, and then 1 year after the initial colonoscopic examination. Fifteen thousand colonoscopies were examined from multiple centers, for which the calculated miss rate for all polyps was 17% and the miss rate for neoplastic polyps 12%. However, the percentage of missed neoplastic polyps greater than 9 mm was only 2%. This report was a retrospective analysis of findings on a repeat colonoscopic examination between 3 and 7 years following the initial colonoscopic examination (assuming that large polyps had been overlooked in the initial colonoscopy), and stated that the overall miss rate for “advanced adenomas” (those larger than 10 mm) was 1.7%. However, this study was not a tandem colonoscopic examination and it was assumed that any lesion over 10 mm in diameter was missed on the original procedure.

A meta-analysis comparing CTC, air contrast barium enema, and colonoscopy found 9 studies that reported segmental unblinding of the CTC findings for the colonoscopist, but used the findings of a single colonoscopy as the factor that determined whether a polyp was or was not present (colonoscopy was the gold standard). Another meta-analysis comparing the accuracy of CTC with colonoscopy reviewed 47 studies in which the findings on CTC were corroborated or not by conventional colonoscopy or by surgery, and found the results were “highly heterogeneous.” A report from Europe compared CTC with segmental unblinding during colonoscopy, but there was no mention of any lesion missed by colonoscopy. In a more recent article, same-day colonoscopy with segmental unblinding was performed, but this report did not reveal how many polyps found on CTC were actually missed by the colonoscopist after the CTC results were revealed. This report on 202 patients did mention that there was one polyp detected on CTC that was missed at initial colonoscopy but found on repeat colonoscopy. Another group also performed segmental unblinding in a prospective same-day CTC and colonoscopy in 311 patients, but there was no mention of any lesion reported on CTC that was not discovered during the subsequent unblinded colonoscopic examination.

A study published in 2008 reported on the results of colonoscopy following a positive CTC examination whereby a polyp or mass was seen that was greater than 9 mm in diameter or at least 3 medium-sized polyps (6–9 mm) were reported. Most patients in this prospective report had colonoscopy within several hours of CTC. Although patients with large polyps typically were seen within several hours of CTC, the timing of colonoscopy had to be at most 30 days after CTC examination in order for the data to be included. In this study, the findings of colonoscopy examination were taken as the standard, and the colonoscopists were told exactly where the lesion was located. There was no attempt at performing a second colonoscopy if the first examination did not reveal an abnormality, and there was a false-positive CTC finding of 5% when the colonoscopy failed to locate a polyp.

In an early multicenter study involving 600 participants, 9 clinical centers were recruited. Colonoscopies were performed with endoscopists blinded to the CTC results, with the CTC finding revealed after the colonoscopist examined each segment of the colon during scope withdrawal. In this study, conventional colonoscopy missed only one 7 mm lesion in the sigmoid colon and 19 lesions that ranged in size from 1 to 5 mm.

Colonoscopy was the reference standard in a 2009 article where CTC was evaluated for the detection of advanced neoplasia in persons at high risk for colon cancer. The investigators noted that “colonoscopy itself may miss some lesions.” In this study, where lesions less than 6 mm in size were reported as negative, a total of 93 cases had a positive CTC but lesions were not found on the subsequent “reference” colonoscopy performed about 3 hours after CTC. Each segment of the bowel was unblinded to the examiner once that area of the colon had been evaluated colonoscopically. In this study, a positive CTC result was recorded if the colonoscopic examination revealed at least one “advanced neoplasia 6 mm or larger” but if no polyp was seen on colonoscopy, the CTC was regarded as a false positive. Ninety-three cases were classified as CTC false positive when colonoscopy did not find a polyp. Blinded colonoscopy missed 2 advanced adenomas, a 13 mm pedunculated polyp in the cecum, and a 18 mm flat lesion in the ascending colon. This article did not state the number of colonoscopies when a lesion seen on CTC was missed on initial colonoscopy but was subsequently found on the second colonoscopic examination.

In a comparison of miss rates on colonoscopy with findings at surgical resection of an index lesion, 16 more lesions were present on the surgical resection specimen in addition to all neoplasms detected at the presurgical colonoscopy. Most polyps were small and only one polyp greater than 1 cm was missed, and that tumor was in the ascending colon. It does seem that comparison of findings with either CT or colonoscopy would be best served by examining a surgical resected specimen to truly ascertain the miss rate of CTC or colonoscopy.

Despite the greater sensitivity for polyp discovery with CTC over DCBE, the US Preventive Services Task Force has not endorsed CTC as a diagnostic procedure for their guidelines on screening recommendations. The recommendation of the US Preventive Services Task Force “concludes that for CT colonography there is insufficient evidence to permit a recommendation for colorectal cancer screening.” This guideline was developed to assess and recommend preventive care services for any patient without signs or symptoms of the target condition.

The most recent guideline on screening for colorectal cancer from the American College of Gastroenterology (ACG) stated that colonoscopy every 10 years, beginning at age 50, is the preferred colorectal cancer screening strategy, but in cases where colonoscopy may not be available or that persons are unwilling to undergo colonoscopy, then CTC every 5 years is an acceptable alternative. Another guideline has been issued by the American Cancer Society, the US Multisociety Task Force on Colorectal Cancer, and the American College of Radiology. This study group does recommend CTC for screening purposes because of the “accumulation of evidence […] the expert panel concludes that there are sufficient data to include CTC as an acceptable option for colorectal cancer screening.”

It is not surprising that lesions may be missed on colonoscopy. There is not any endoscopist who performs colonoscopy who has not seen a polyp, lost sight of it while waiting for a snare or biopsy forceps, and then needed to search for it again. Similarly, polyps found during intubation and not removed may be very difficult to locate during removal of the instrument. Intubation of the colon is characteristically performed rather rapidly for several reasons, one of which is to minimize patient discomfort by shortening the examination. Another is to reduce spasm in the colon that will occur if the procedure is prolonged, and to avoid the necessity of overdistending the right colon with a slow intubation. Because of this, the colonoscopic examination is best performed during withdrawal of the instrument, which must be carefully controlled. This paradigm, performing a rapid insertion with little or no emphasis on inspection, followed by inspection during the withdrawal phase, has not been scientifically proven to be an optimal approach for achieving maximum detection of adenomas or cancer. Because the extubation is the portion of the procedure during which time most adenomas are found, careful withdrawal is of utmost importance. In 2006, a combined task force of the ACG and the American Society for Gastrointestinal Endoscopy, in a combined statement, recommended that the withdrawal phase of colonoscopy should be an average of 6 minutes in duration. A private practice group scrutinized their data and found that there was a strong correlation with withdrawal time and adenoma detection rate. In this study, Barclay and colleagues reported that colonoscopists with an average withdrawal time of more than 6 minutes detected adenomas in 6.4% of screened patients compared with a 2.6% prevalence in colonoscopies performed by endoscopists whose withdrawal times averaged less than 6 minutes. The Mayo Clinic also validated the 6-minute withdrawal target as separating high from low adenoma detectors. During the 6-minute withdrawal, the colonoscopist must make an assessment of each fold and try to visualize the area behind folds and on the inner aspect of angulations in the colon. The usual technique during withdrawal around a fold is rather complex, and requires skill and dexterity : (A) withdraw air, which shortens the colon and moves the colonoscope tip proximal to the fold; (B) the tip is angulated toward the fold and withdrawn, with the colonoscope tip deflected in the direction of the fold; (C) this pulls on the fold and bends the fold toward the examiner, permitting visualization of the space behind the fold.

Whenever a fold or flexure is passed and a careful examination of its proximal aspect cannot be achieved, reinsertion, flexion of the tip, and repeat withdrawal is necessary. An attempt should be made to avoid a “red out” whereby nothing is seen when the tip is deflected behind a fold. The angle of deflection is controlled with the left thumb on the major up/down control knob as the instrument is withdrawn, moving the tip toward the fold, while permitting visualization of its hidden portion. A retroflexion should routinely be performed in the rectum. It would seem that retroflexion of the instrument during the withdrawal phase at any location in the colon would be a worthwhile adjunct, but an article has stated that retroflexion in the right colon was not able to visualize any greater amount of any additional pathology than seen with straight end-on colonoscopy. Various techniques have been attempted to increase the ability to see portions of mucosa hidden during the withdrawal phase. One of these techniques is to use a cap on the end of the instrument while another is to use a wide-angle instrument. Studies have not identified improved overall adenoma detection using these devices. Pickhardt and colleagues, in comparing colonoscopy with CTC, use a computer-simulated graphic representation of the area behind folds that cannot be seen with the straight end-on colonoscopic view during withdrawal of the instrument. Lieberman commented in an editorial that “the data on colonoscopy accuracy [is] a humble reminder of the limitation of colonoscopy; nevertheless it remains the pre-eminent test for diagnosing and treating colonic neoplasia.”

A more recent addition to the quest for a more thorough colonoscopic examination has been a mini-endoscope that permits both antegrade and retrograde visualization simultaneously during colonoscopic withdrawal. This device is called the Third Eye Retroscope (TER). When placed through the instrument channel of a standard colonoscope it flexes 180° as it emerges from the tip and extends into the lumen ( Fig. 1 ). The device carries a light source and a viewing chip. The light from the retroverted instrument illuminates the areas distal to the colonoscope tip, permitting the device to visualize the proximal portions of folds and the valleys in between these folds as the colonoscope simultaneously is looking forward ( Fig. 2 ). The TER has shown an increased detection rate for polyps and adenomas. It has been suggested that this instrument is “one of the most promising devices for improvement of mucosal exposure during colonoscopy” ( Fig. 3 ). Preliminary reports of the TER have been published. A prospective multicenter study by 14 endoscopists at 8 sites studied 249 patients who presented for screening or surveillance colonoscopy. Following cecal intubation, the disposable TER was inserted through the instrument channel of the colonoscope. During withdrawal, the forward and retrograde video images were observed side by side simultaneously on a wide-screen monitor. The number and sizes of lesions detected with the standard colonoscope were recorded, as were the number and sizes of lesions found that were first detected with the TER, but not with the forward viewing colonoscope. ( Fig. 4 ) In these subjects, 257 polyps including 136 adenomas were identified with the colonoscope. The TER detected 34 additional polyps with (a 13.2% increase over standard forward viewing colonoscopy). These 34 polyps included 15 adenomas (an 11% increase over those seen with the standard forward viewing colonoscope). The additional detection rate with the TER compared with standard colonoscopy for any lesion greater than or equal to 10 mm was 30.8% for all polyps and 33.3% for adenomas. Every polyp detected with the TER was subsequently located with the colonoscope and removed. Another prospective multicenter study involving 17 investigators and 298 patients at 9 sites had a study design that was similar except that the endoscopists were initially naïve to the Third Eye device and were followed through the “learning curve” over 20 procedures. Their overall additional detection rates for the Third Eye compared to the colonoscope alone were 14.8% for all polyps and 16.0% for adenomas. For procedures performed after each endoscopist had completed 15 cases using the device, mean additional detection rates with the Third Eye were 17.0% for all polyps and 25.0% for adenomas. As in the prior multicenter study , additional adenoma detection rates with the Third Eye were greater for larger lesions compared to smaller lesions, suggesting that at least some of the lesions that were hidden from the view of the colonoscope might have been missed during previous exams.