New high-resolution colonoscopes and filter technologies are allowing us to visualize more lesions and better characterize lesions within the gastrointestinal tract. In light of recent findings that flat and serrated lesions are more likely to contain invasive cancer and that even small lesions (5–10 mm) may contain advanced histology, detecting these lesions earlier with improved optical technologies may help decrease the rate of interval cancers after colonoscopy. With the limited accuracy of white-light colonoscopy (59%–84%) in distinguishing non-neoplastic lesions from neoplastic lesions, these new technologies can help us improve our abilities to risk stratify patients and determine more precise surveillance intervals.
Recent concerns that the protective effect of colonoscopy is lower than previously thought have shifted attention to improving the precision of colonoscopy. Although efforts at optimizing the quality of colonoscopies are ongoing, a pooled adenoma detection miss rate of 22% (15%–32%) in tandem colonoscopy studies by experienced endoscopists and increasing recognition of nonpolypoid flat premalignant lesions, have lead to accelerated interest in technological developments to improve adenoma detection and assist in the characterization of premalignant lesions.
The prognosis for patients with malignancies of the lower gastrointestinal (GI) tract is strictly dependent on early detection of premalignant and malignant lesions. Three steps are important for a proper endoscopic diagnosis: recognition, characterization, and confirmation. Computer and chip technologies have faced major technological advances that can improve endoscopic diagnostics. High-resolution or high-definition (HD) endoscopes can improve recognition or detection of lesions of the colon. Chromoendoscopy or filter-aided colonoscopy (virtual chromoendoscopy) can enhance characterization of lesion morphology and surface architecture to better predict histology. Finally, conventional or in vivo histology (confocal laser endomicroscopy) can provide histologic confirmation to define whether neoplastic changes are present or not.
The resolution of an endoscopic image is different from magnification, and is defined as the ability to distinguish between 2 points that are close together. High-resolution imaging improves the ability to discriminate details while magnification enlarges the image. In digital video imaging, resolution is a function of pixel density. Through high-pixel density charged-coupled devices (CCD), high-resolution endoscopes provide slightly magnified views of the GI tract but with greater mucosal detail. Magnification endoscopy uses a movable lens controlled by the endoscopist to vary the degree of magnification, which ranges from 1.5× to 150×. Newly designed magnification endoscopes provide both high-resolution and magnification features.
More recently, HD endoscopes are available. CCDs convert light information into an electronic signal. This signal is processed into an image via the video processor. The standard analog broadcasting systems (Phase Alternate Line [PAL] and National Television System Committee [NTSC]) generate approximately 480 to 576 scanning lines on a screen. Now, the new HD endoscopes can generate up to 1080 scanning lines on a screen, which further increases the resolution. Surface analysis on distinct lesions can be performed even before magnification.
High-Definition Colonoscopy for the Detection of Colorectal Neoplasia
Theoretically, the effectiveness of endoscopy should depend on the resolution. The better the image quality, the more likely it should be to obtain the right diagnosis. Rex and colleagues called for further evaluation of high-definition endoscopy for adenoma detection when they noted a high number of subjects with greater than or equal to 1 adenoma regardless of whether they were randomized to withdrawal with white light using HD endoscopes or narrow band imaging (67% vs 65%; p = ns). It was unclear whether high adenoma detection rates in this study were caused by improved visualization from HD endoscopes or simply precise withdrawal technique from an experienced operator. However, the few dedicated studies subsequently evaluating the effectiveness of HD endoscopy over standard resolution (SR) endoscopy in adenoma detection have been conflicting. Table 1 summarizes the recent published data.
|East et al||2008||Prospective, cohort study||130||Subjects with at least 1 adenoma||71%||60%||Not significant|
|Pellisé et al||2008||Prospective, randomized controlled||620||Mean number of adenomas per subject||0.43||0.45||Not significant|
|Tribonias et al||2010||Prospective randomized||390||Polyp detection rate per subject||1.76||1.31||P = .03 |
(no difference for adenoma detection)
|Burke et al||2010||Retrospective, comparative cohort study||852||Polyp detection rate||39.9%||36.9%||Not significant|
|Buchner et al||2010||Retrospective study||2430||Adenoma detection rate||28.8%||24.3%||P = .012|
|Hoffman et al||2010||Prospective, randomized controlled||200||Subjects with at least 1 adenoma||38%||13%||P <.0001|
Several studies suggest that high-definition colonoscopy is not significantly better than SR colonoscopy at improving adenoma detection. East and colleagues compared adenoma detection rates using HD colonoscopy versus SR colonoscopy in a prospective cohort study of 130 subjects where colonoscopy was performed by 1 experienced endoscopist using optimal withdrawal techniques. This study revealed no significant difference in adenoma detection rates (71% vs 60%; P = .20), number of proximal hyperplastic polyps (31 vs 23; P = .35), and number of overall adenomas detected (93 vs 88; P = .12) between the two technologies, respectively. However, for small (<6 mm), nonflat adenomas, a significantly higher number of lesions were detected with HD ( P = .03). They concluded that high-quality withdrawal techniques in both groups may have been the most important factor contributing to high adenoma detection rates in both SR and HD colonoscopy.
In a randomized, controlled trial including 620 subjects, Pellisé and colleagues also investigated whether HD colonoscopy was more effective in adenoma detection than SR colonoscopy by a group of 7 full-time gastroenterologists who spent more than 50% of their time performing procedures. Both techniques detected a similar number of adenomas on a per-subject analysis (HD 0.43 vs SR 0.45; p = ns), with no differences in the type of lesions detected, distribution of lesions along the colon, the degree of dysplasia, or morphology of the adenomas. Even in standard practice conditions, HD colonoscopy did not detect significantly more colorectal neoplasia than SR colonoscopy.
In a smaller prospective study, Tribonias and colleagues did note a significantly increased polyp detection rate per subject in the HD versus SR colonoscopy group (1.76 vs 1.31; P = .03). However, the adenoma detection rate did not differ significantly in the two groups. Similarly, Burke and colleagues powered their comparative retrospective cohort study of 852 subjects to detect a 10% difference in polyp detection between HD and SR colonoscopy and yielded negative results. However, they showed that HD improved the detection of subjects with greater than or equal to 3–6-mm or smaller adenomas ( P = .050).
Two more recent studies have shown that HD colonoscopy can improve overall adenoma detection. Although retrospective, the study by Buchner and colleagues is the largest to date with 2430 nonselected subjects in a general practice setting by various endoscopists. Not only was there a statistically significant increase in polyp detection (HD 42.2% vs SD 37.8%; P = .026) but also adenoma detection, (HD 28.8% vs SR 24.3%; P = .012) even after adjusting for potential confounding factors. Once again, the highest polyp detection with HD occurred in polyps 0 to 5 mm in size, (HD 32.3% vs SR 27.2%; P = .009). In a smaller prospective study, Hoffman and colleagues also showed that HD colonoscopy is superior in adenoma detection and in identifying significantly more patients with at least 1 adenoma.
HD deserves further study to clarify its clinical value. Many of the studies have suggested that there is increased detection of diminutive adenomas (1–5 mm) with this technology. Because diminutive lesions can harbor unfavorable histology, the detection and removal of these lesions may be clinically important. However, the value of detecting and removing diminutive adenomas (1–5 mm) still remains unclear. Regardless of conflicting data, the transition from standard resolution to high-definition endoscopy is likely inevitable; endoscopists prefer high definition because they simply see better and there is no learning curve for its use.
Chromoendoscopy or tissue staining is an old endoscopic technique that has been used for decades. It involves the topical application of stains or pigments to improve localization, characterization, or diagnosis of a lesion. It is a useful adjunct to endoscopy; the contrast between normally stained and abnormally stained epithelium enables the endoscopist to make a diagnosis or to direct biopsies based on a specific reaction or enhancement of surface morphology.
The technique for staining is simple and easy to learn. Chromoendoscopy can be done in an untargeted fashion of the whole colon (panchromoendoscopy) or directed toward a specific lesion (targeted staining). While spraying dyes in the colon in an untargeted fashion, the endoscopist needs to direct the endoscope and catheter tip toward the colorectal mucosa and use a combination of rotational clockwise-counter clockwise movements with simultaneous withdrawal of the endoscope tip. The movements are necessary to achieve an even spread of the dye on the mucosa.
Contrast dyes, which coat the colonic mucosa, include 0.1% to 0.4% indigo carmine. This dye has been used for most studies of chromoendoscopy in the colon. The dye is not absorbed but simply highlights surface topography by pooling in grooves caused by mucosal lesions. In addition, 0.1% methylene blue has been used in colonoscopy and is an absorptive dye actively taken up by normal epithelial cells in the colon and small intestine. There has been some suggestion that methylene blue can cause oxidative DNA damage in cells exposed to white light (WL) during chromoendoscopy. However, a recent study did not show an increase in cancers in subjects who underwent chromoendoscopy with methylene blue.
Chromoendoscopy for improved adenoma detection
Although individual authors drew different conclusions regarding the effectiveness of chromoendoscopy over conventional colonoscopic examinations for adenoma detection, a recent Cochrane review pooled the overall experience with chromoendoscopy for adenoma detection from 4 prospective, randomized trials in subjects without inflammatory bowel disease or polyposis syndromes ( Table 2 ). Despite nonsignificant adenoma detection rates in 3 of the 4 studies, when pooled together, the studies revealed that chromoendoscopy yields more subjects with at least 1 neoplastic lesion (odds ratio [OR]: 1.61 [95% confidence interval (CI) 1.24–2.09]) and more subjects with 3 or more neoplastic lesions (OR 2.55 [95% CI 1.49–4.36]) than standard colonoscopy. The review authors concluded that there is
strong evidence that chromoendoscopy enhances the detection of neoplasia in the colon and rectum. Patients with neoplastic polyps, particularly those with multiple polyps, are at increased risk of developing colorectal cancer. Such lesions, which presumably would be missed with conventional colonoscopy, could contribute to the interval cancer numbers on any surveillance program.
|Brooker et al||2002||Randomized study: 1 procedure||259||Total number of adenomas||125||49||Not significant|
|Hurlstone et al||2004||Randomized study: 1 procedure||260||Subjects with ≥3 adenomas||13||4||P <.01|
|Le Rhun et al||2006||Randomized study: 1 procedure||198||Total number of adenomas per subject||0.6||0.5 (HRC)||Not significant|
|Lapalus et al||2006||Tandem colonoscopy; randomized second scope||292||Subjects with ≥1 adenoma||40%||36% (HRC)||Not significant|
|Brown et al |
|2007||All prospective randomized studies with full chromoendoscopic evaluation of colon||1009||a) Subjects with ≥1 adenoma |
b) Subjects with ≥3 adenoma
|a) OR in favor of chromo: 1.6 |
b) OR in favor of chromo: 2.55
|95% CI: 1.24– 2.09 |
95% CI: 1.49–4.36
|Stoffel et al||2008||Tandem colonoscopy; randomized second scope||50||Adenomas per subject on second examination||0.7||0.2||P <.01|
A more recent multicenter study conducted in the United States randomized 50 subjects with a history of colorectal cancer or adenomas to tandem colonoscopy with the second examination as standard colonoscopy or chromoendoscopy. This study found that the second chromoendoscopy detected adenomas in more subjects than did intensive 20-minute inspection with a second standard colonoscopy (44% vs 17%), changing management in 27% of subjects who would have been classified as free of adenomas. Similar to the Lapalus study, the majority of adenomas detected were significantly smaller (2.66 mm ± 0.97 mm) and right sided.
Despite increasing polyp and adenoma detection with chromoendoscopy, studies have repeatedly demonstrated a longer withdrawal time. Although withdrawal techniques differed in the 4 studies included in the Cochrane review, withdrawal time was longer in the chromoendoscopy group (3–75 minutes) than with the control group (2–60 minutes). In the Stoffel study, despite the 20-minute intensive inspection in the standard colonoscopy arm, chromoendoscopy still required more time than intensive inspection, (36.9 vs 27.3 minutes; P <.01).
Several studies have attempted different methods of staining to improve procedure time and to maintain high adenoma detection rates with chromoendoscopy. As attention has focused on flat or diminutive, right-sided lesions, Park and colleagues randomized 316 consecutive subjects to a repeat examination of the ascending colon and cecum only with a second colonoscopy or chromoendoscopy with indigo carmine. During the second diagnostic intubation, 17.4% of subjects in the chromoendoscopy group and 5.3% in the standard colonoscopy group had at least 1 adenoma ( P <.001). There was no statistically significant difference in procedural time between the two groups as chromoendoscopy only occurred in the ascending colon and cecum and standard colonoscopy was fixed to at least 120 seconds of inspection time. If the clinically significant flat adenomas are mainly right sided, this method of focusing examination to the ascending colon and cecum could increase diagnostic yield of adenomas without adding the procedural burden of staining the entire colon.
However, Hurlstone and colleagues randomized 260 subjects to panchromoendoscopy or targeted chromoendoscopy and discovered that although extubation times did not significantly differ between the groups (Chromo [median 17 minutes] vs targeted [median 15 minutes]), in the panchromoendoscopy group, more adenomas were detected ( P <.05), more diminutive (<4mm) adenomas were detected overall ( P = .03), and most importantly, more flat and diminutive adenomas were detected in the right colon ( P <.05). This study suggests that panchromoendoscopy is better for improved adenoma detection than targeted chromoendoscopy. Nevertheless, chromoendoscopy is still regarded as cumbersome and time intensive and has not been widely incorporated into everyday practice.
Diminutive adenomas contributed most to the higher adenoma detection with chromoendoscopy in the majority of the studies. As previously mentioned, the significance of these diminutive adenomas require further investigation, but if recent data suggesting that 7% to 15% of small adenomas (5–10mm) show advanced histology, further optical enhancements, such as chromoendoscopy, should be routinely adopted in clinical practice.
Chromoendoscopy for improved characterization
In addition to increased detection, chromoendoscopy has been evaluated for its characterization abilities. Surface analysis of stained colorectal lesions was a new optical impression for the endoscopists in the 1990s. First, Kudo and colleagues described that some of the regular staining patterns are often seen in hyperplastic polyps or normal mucosa; whereas, unstructured surface architecture was associated with malignancy. Also, adenomas can be classified better (tubular vs villous) upon detailed inspection. This experience has lead to a categorization of the different staining patterns in the colon: The so-called pit-pattern classification differentiated 5 types and several subtypes. Types 1 and 2 are staining patterns predicting non-neoplastic lesions; whereas, types 3 to 5 are predicting neoplastic lesions. With the help of this classification, the endoscopist may predict histology with good accuracy.
Axelrad and colleagues first reported a sensitivity and specificity of chromoendoscopy of 95% and 93%, respectively, with a diagnostic accuracy of 81%, but the ability of chromoendoscopy to differentiate between adenomatous polyps from nonadenomatous polyps have varied considerably in several large studies using indigo carmine with sensitivity and specificity between 82% to 98% and 52% to 95%, respectively ( Table 3 ).
|Author||Year||Design||N||Outcome||Sensitivity (%)||Specificity (%)||Overall Accuracy (%)|
|Axelrad et al||1996||Prospective study; lesion analysis||36||High-resolution, magnifying Chromo polyp analysis||95||93||81|
|Togashi et al||1999||Prospective study; lesion analysis||1280||Magnifying chromo polyp analysis||92.0||73.3||88.4|
|Tung et al||2001||Prospective study; lesion analysis||141||Magnifying chromo polyp analysis||93.8||64.6||80.1|
|Eisen et al||2002||Prospective, multicenter trial; lesion analysis||299||High-resolution, Chromo polyp analysis||82||82||88|
|Konishi et al||2003||Prospective, randomized study||660||Standard chromo vs Magnifying chromo||SC: 90 |
|SC: 61 |
|SC: 68 |
|Fu et al||2004||Prospective study; lesion analysis||122||Conventional colonoscopy vs Chromo vs Magnifying chromo||CC: 88.8 |
|CC: 93.1 |
|CC: 84.0 |
|Su et al||2004||Prospective study; lesion analysis||230||Magnifying chromo polyp analysis||95.1||86.8||91.9|
Magnifying chromoendoscopy can increase the diagnostic accuracy even further. Konishi and colleagues randomized 660 subjects to standard chromoendoscopy with 0.2% indigo carmine and magnifying chromoendoscopy in a prospective fashion to determine the surface staining pattern (pit-pattern classification) and predict the malignant potential of lesions. Magnifying endoscopes significantly improved diagnostic accuracy (92%) as compared with chromoendoscopy with standard video endoscopes (68%).
Although interobserver and intraobserver variability among experienced endoscopists for determining pit pattern is good (κ = 0.72; κ = 0.81, respectively), chromoendoscopy is still not widely used in clinical practice for the characterization of lesions. Once again, the time-intensive nature of chromoendoscopy may limit its use, but the question remains whether more readily available high-definition or high-resolution endoscopes may help with determining pit patterns without the use of magnifying endoscopes. In one study, 150 subjects undergoing high-resolution screening colonoscopy were evaluated for polyps less than 5mm in the rectum and sigmoid colon. The sensitivity and specificity for predicting non-neoplastic polyps only minimally improved from 93% to 94% and 60% to 64%, respectively, with the addition of indigo carmine. As high-definition and high-resolution colonoscopes are more convenient to use and more readily available, studies should evaluate whether this technology can help characterize lesions as well as magnifying chromoendoscopy.
Chromoendoscopy in high-risk populations
Hereditary nonpolyposis colorectal cancer
Given the fast progression of hereditary nonpolyposis colorectal cancer (HNPCC) syndrome lesions to cancer, with data showing 4 out of 11 interval cancers within 3.5 years of normal colonoscopic assessment, patients with HNPCC syndrome might benefit from enhanced endoscopic imaging to help detect flat and depressed lesions. Three tandem colonoscopy studies sought to discover the potentially added benefit of chromoendoscopy in this high-risk patient population. Among 25 asymptomatic subjects with HNPCC in Hurlstone’s study, targeted chromoendoscopy identified 24 lesions in 13 subjects, but panchromoendoscopy helped detect 52 more lesions in 16 subjects, with more adenomas detected in the panchromoendoscopy group ( P = .001). Lecomte and colleagues also found significantly more adenomas on the second chromoendoscopy examination of the proximal colon than on the first examination. Stoffel and colleagues went on further to say that the second examination itself with white light lead to higher adenoma detection with minimal contribution from chromoendoscopy. Because the study was small and not powered to detect a difference in adenoma detection between intensive inspection on second examination and chromoendoscopy, future multicenter trials may need to look into adenoma detection rates with not only chromoendoscopy but other enhancing imaging modalities.
Patients with ulcerative colitis (UC) have a significantly higher risk for the development of colitis-associated colorectal cancer, with a reported cumulative incidence rate of colorectal cancer that is 2.5% at 20 years, 7.6% at 30 years, and 10.8% at 40 years, with higher rates if patients have primary sclerosing cholangitis. Accepted guidelines call for 2 to 4 random biopsies every 10 cm of the colon to look for dysplastic changes. Despite this, 16 of 30 cancers are interval cancers for patients already in surveillance programs. With growing knowledge that dysplastic changes are subtle yet visible, we need new methods to improve our targeted biopsies.
A large study using magnifying chromoendoscopy in 886 subjects with UC found that dysplasia and early cancer were characterized by granular or nodular protruding mucosa or by lowly protruding or flat mucosa, often associated with redness. Dye-spraying endoscopy was useful for detection of all lesions with intraepithelial neoplasias characterized by a staining pattern III-V ( Fig. 1 ). Dysplasia was never found in normal-looking mucosa after staining. Thus, enhanced imaging could potentially improve the detection of these visible lesions. Random biopsies are still widely performed but cannot eradicate the fear of overlooked cancers.
In 2003, the first randomized, controlled trial was published to test whether chromo and magnifying endoscopy might facilitate early detection of intraepithelial neoplasia in patients with ulcerative colitis. A total of 165 subjects with long-standing ulcerative colitis were randomized at a 1:1 ratio to undergo conventional colonoscopy or colonoscopy with chromoendoscopy using 0.1% methylene blue. Lesions in the colon were evaluated according to a modified pit-pattern classification, (pit-pattern I-II: endoscopic prediction non-neoplastic; pit-pattern III-V: endoscopic prediction neoplastic). In the chromoendoscopy group, there was a significantly better correlation between endoscopic assessment of degree and extent of colonic inflammation and histopathologic findings compared with the conventional colonoscopy group. More targeted biopsies were possible, and significantly more intraepithelial neoplasias were detected in the chromoendoscopy group (32 vs 10). Furthermore, using the modified pit-pattern classification and chromoendoscopy, both the sensitivity and specificity for differentiation between non-neoplastic and neoplastic lesions were 93%. In a subsequent study, chromoendoscopy with confocal endomicroscopy increased the diagnostic yield of intraepithelial neoplasia 4.75 fold as compared with conventional colonoscopy and biopsy techniques.
Hurlstone detected the intraepithelial neoplasia detection rate from targeted biopsies to be 8% (49 of 644) versus 0.16% (20 of 12,850) from nontargeted biopsies, suggesting that the random biopsies are low yield. Among 350 subjects with UC, magnifying chromoendoscopy yielded 69 lesions compared with only 24 with standard colonoscopy. Rutter and colleagues similarly found no dysplastic tissue in 2904 nontargeted biopsies, and detected 9 dysplastic lesions in 157 targeted biopsies using chromoendoscopy. More recently, Marion and colleagues found that with targeted chromoendoscopy, increased number of biopsies confirmed low- and high-grade dysplastic tissue [16 and 1], respectively, than random biopsies [3 and 0]; (p = 0.001) and targeted non-dye spray colonoscopy [8 and 1]; ( p = 0.057).
Taken together, these data suggest that targeted biopsies after dye staining will replace random biopsies in the future. Magnifying chromoendoscopy is a valid tool for better endoscopic detection of intraepithelial neoplasia in patients with long-standing ulcerative colitis and an increased ability to differentiate non-neoplastic from neoplastic lesions. A recent Medical Position Statement of the American Gastroenterological Association endorsed the use of chromoendoscopy in the surveillance of patients with UC for physicians who have expertise with this technique. The European consensus guidelines even adopted the concept of smart biopsies of lesions detected and analyzed by chromoendoscopy as an alternative to untargeted quadrant biopsies of apparently normal mucosa after chromoendoscopy. Future studies should longitudinally evaluate if improved outcomes will result from improved dysplasia detection in patients with ulcerative colitis.
Despite increased dysplasia detection in higher-risk patient populations with chromoendoscopy and increased adenoma and polyp detection rates for the general population, it is not widely used in clinical practice because of the cumbersome nature of the dye spray. As a result, digital modalities have developed to provide enhanced chromoendoscopy images without the use of dye. Conventional white-light endoscopy uses the full, visible wave-length range to produce a red-green-blue image. New filter technologies can narrow red-green-blue bands (ie, narrow band imaging [NBI]) to enhance microvessel architecture or use adaptive image postprocessing algorithms (filtering and logic) to segment and extract image irregularities (ie, i-scan [Pentax Medical Company, Montvale, NJ, USA]; Fujinon Intelligent Color Enhancement system [FICE], [Fujinon, Inc, Wayne, NJ, USA]) ( Fig. 2 ). This technology can modulate different forms of enhancement, which leads to an accentuation of the vasculature, the surface architecture, or the pattern visualization.