Although removal of adenomatous polyps has been shown to decrease the risk of colon cancer, distal hyperplastic polyps are thought to not have malignant potential. Most polyps detected during colonoscopy are diminutive (≤5 mm) and rarely harbor advanced histology, such as high-grade dysplasia or cancer. Therefore, predicting histology in real-time during colonoscopy can potentially decrease the enormous expenditure that ensues from universal histopathologic evaluation of polyps, and several novel imaging technologies have been developed and tested over the past decade for this purpose. Of these different technologies, electronic chromoendoscopy seems to strike a fair balance between accuracy, feasibility, and cost.
Key points
- •
Real-time polyp histology prediction during colonoscopy is feasible and can result in significant cost savings.
- •
Several imaging technologies have been studied and most have shown good accuracy in polyp histology prediction.
- •
Electronic chromoendoscopy is a practical and easy to use technology that is accurate for real-time histology prediction of polyps.
Introduction
Colonoscopy is the favored modality for screening and prevention of colorectal cancer in the United States. Removal of adenomatous polyps during colonoscopy can prevent the development of colorectal cancer, because most cancers arise from adenomatous polyps following the adenoma-carcinoma pathway. Another type of colon polyp are hyperplastic polyps, which are in general not considered to be premalignant, especially those present in the distal colon and 5 mm or less in size. Most polyps (>80%) detected during colonoscopy are diminutive (ie, ≤5 mm). Current practice is to remove all polyps detected during colonoscopy, irrespective of their size, and send them to pathology for evaluation. Because diminutive polyps rarely harbor any advanced histology, such as high-grade dysplasia or cancer, histopathologic evaluation of these lesions determines whether they are adenomatous, which then dictates the postpolypectomy surveillance intervals. This practice entails a huge cost burden to the health care system, with limited clinical benefits in return. If the histology of diminutive polyps can be predicted real-time during colonoscopy by the endoscopist, then the cost of histology can potentially be avoided (“resect and discard”).
Moreover, although histopathologic diagnosis is considered the gold standard, another advantage of predicting polyp histology relates to diminutive hyperplastic polyps. Because these are not considered to have malignant potential, their removal and routine pathologic evaluation are seemingly wasteful and time-consuming endeavors. If these can be accurately characterized during colonoscopy, then they can potentially be left behind (“do not resect”), thereby saving further costs and also decreasing the risks associated with polypectomy. Cost-savings estimates of the “predict, resect, and discard” strategy have ranged from $33 million to $1 billion annually in the United States.
Several novel imaging modalities have been developed during the past decade that have expanded the scope of colonoscopy and enabled the endoscopist to predict polyp histology. These imaging systems can be broadly divided into large-field and small-field technologies depending on the area of mucosa that is imaged. Large-field technologies include high-definition white light, chromoendoscopy, electronic chromoendoscopy, and autofluorescence. Small-field technologies include confocal endomicroscopy and endocytoscopy. This article reviews each technology as it relates to predicting polyp histology.
Introduction
Colonoscopy is the favored modality for screening and prevention of colorectal cancer in the United States. Removal of adenomatous polyps during colonoscopy can prevent the development of colorectal cancer, because most cancers arise from adenomatous polyps following the adenoma-carcinoma pathway. Another type of colon polyp are hyperplastic polyps, which are in general not considered to be premalignant, especially those present in the distal colon and 5 mm or less in size. Most polyps (>80%) detected during colonoscopy are diminutive (ie, ≤5 mm). Current practice is to remove all polyps detected during colonoscopy, irrespective of their size, and send them to pathology for evaluation. Because diminutive polyps rarely harbor any advanced histology, such as high-grade dysplasia or cancer, histopathologic evaluation of these lesions determines whether they are adenomatous, which then dictates the postpolypectomy surveillance intervals. This practice entails a huge cost burden to the health care system, with limited clinical benefits in return. If the histology of diminutive polyps can be predicted real-time during colonoscopy by the endoscopist, then the cost of histology can potentially be avoided (“resect and discard”).
Moreover, although histopathologic diagnosis is considered the gold standard, another advantage of predicting polyp histology relates to diminutive hyperplastic polyps. Because these are not considered to have malignant potential, their removal and routine pathologic evaluation are seemingly wasteful and time-consuming endeavors. If these can be accurately characterized during colonoscopy, then they can potentially be left behind (“do not resect”), thereby saving further costs and also decreasing the risks associated with polypectomy. Cost-savings estimates of the “predict, resect, and discard” strategy have ranged from $33 million to $1 billion annually in the United States.
Several novel imaging modalities have been developed during the past decade that have expanded the scope of colonoscopy and enabled the endoscopist to predict polyp histology. These imaging systems can be broadly divided into large-field and small-field technologies depending on the area of mucosa that is imaged. Large-field technologies include high-definition white light, chromoendoscopy, electronic chromoendoscopy, and autofluorescence. Small-field technologies include confocal endomicroscopy and endocytoscopy. This article reviews each technology as it relates to predicting polyp histology.
High-definition white light endoscopy
Technology
The older-generation white light endoscopy used a charge-coupled device (CCD) chip with an average of 300,000 pixels. Over the past decade, technology has evolved, with advancements in miniaturization and a specialized design of the CCD chip. CCD chips now have a 3-fold higher pixel density than standard-definition white light endoscopy (ie, >1 million pixels). This resolution, along with a high-definition video processor and a high-definition monitor, produces a high-definition image with 1080 effective scan lines. Although some data suggest that high-definition white light (HD-WL) endoscopy may improve the adenoma detection rate, this improvement in technology has not had a significant impact on the ability to characterize the histology of polyps.
Performance
A randomized controlled study reported no difference in the performance between standard-definition and HD-WL endoscopy in characterizing the histology of 293 consecutive polyps smaller than 10 mm, as measured by sensitivity (76% vs 76%; P = .96), specificity (59% vs 67%; P = .44), and accuracy (70% vs 73%; P = .6). Another prospective, randomized controlled trial showed slightly superior accuracy of HD-WL compared with standard-definition endoscopy in predicting adenomatous polyps (73.2% versus 68.5%, P <.0001). A third prospective study compared HD-WL endoscopy and narrow band imaging (NBI) in the prediction of histology of 236 polyps in 100 patients. HD-WL endoscopy had a significantly lower sensitivity (38% vs 96%; P <.0001) and lower accuracy (61% vs 93%; P <.0001) than NBI in distinguishing adenomas from hyperplastic polyps.
Chromoendoscopy
Technique
Chromoendoscopy involves the spraying of dyes, such as methylene blue, cresyl violet, and indigo carmine, using a spray catheter. Methylene blue and cresyl violet stain the surface of a lesion by being actively absorbed and interacting with cell constituents. Contrast dyes such as indigo carmine are not absorbed by the mucosa and pool in the pits and mucosal crevices on the surface of polyps. These dyes can highlight different patterns on the surface of polyps called pit patterns . These patterns were described by Kudo and colleagues using magnification endoscopy and have been shown to accurately differentiate hyperplastic and adenomatous polyps ( Fig. 1 , Table 1 ).
| Type | Description | Most Likely Histology | Neoplastic/Nonneoplastic |
|---|---|---|---|
| I | Round crypts | Normal mucosa | Nonneoplastic |
| II | Regular wider or stellar crypts | Hyperplastic polyp | Nonneoplastic |
| III-L | Elongated or roundish crypts | Adenoma | Neoplastic |
| III-s | Tubular or roundish pits smaller than the normal crypts | Adenoma | Neoplastic |
| IV | Branch-like or gyrus-like crypts | Adenoma | Neoplastic |
| V-i | Irregular crypts | Superficial invasive/deep invasive cancer | Neoplastic |
| V-n | Non-structural crypts | Deep submucosal invasive cancer | Neoplastic |
Performance
A prospective study of 122 patients evaluated 206 polyps less than 10 mm using conventional colonoscopy, chromoendoscopy with indigo carmine, and chromoendoscopy with indigo carmine and magnification. The predicted histology was compared with the gold standard (ie, histopathologic diagnosis of endoscopically resected polyps). The overall sensitivity, specificity, and accuracy for distinguishing between neoplastic and nonneoplastic lesions using conventional colonoscopy were 88.8%, 67.4%, and 84.0%, respectively, although these were 93.1%, 76.1%, and 89.3%, respectively, for chromoendoscopy without magnification, and 96.3%, 93.5%, and 95.6%, respectively, for chromoendoscopy with magnification. Chromoendoscopy using indigo carmine with magnification was significantly superior in distinguishing neoplastic from nonneoplastic lesions, compared with conventional colonoscopy ( P <.0001) and chromoendoscopy without magnification ( P = .0152). However, another study in 141 patients evaluating 175 polyps, of which 161 were less than 10 mm, showed somewhat lower accuracy in histologic characterization using chromoendoscopy with magnification. The overall sensitivity, specificity, and diagnostic accuracy for distinguishing between neoplastic and nonneoplastic lesions were 93.8%, 64.6%, and 80.1%, respectively. Several other studies using chromoendoscopy with magnification have shown diagnostic accuracies between 80% and 96% for distinguishing neoplastic from nonneoplastic lesions.
Feasibility
Chromoendoscopy has not been adopted widely in the western countries because it is perceived to be labor-intensive, cumbersome, and messy. It also increases the duration and cost of the procedure because of the use of dye and spray catheter. Magnification endoscopes were used to describe the pit patterns by Kudo and colleagues, and these are not available in the United States. In addition, concerns exist regarding in vitro data showing that methylene blue can cause single-strand DNA breaks in epithelial cells.
Narrow band imaging
Technology
NBI is by far the most extensively studied of the 3 electronic chromoendoscopy systems. In white light endoscopy, the entire spectrum of visible light (400–700 nm) is used, and therefore the mucosa is seen in its natural color. NBI uses optical filter in the endoscopy system to transmit an increased proportion of blue light (415 nm) and a decreased proportion of red light. It narrows the white light spectrum into 2 different wavelengths: blue (390–445 nm) and green (530–550 nm) light. These wavelengths correspond to the maximum or peak absorption spectrum of hemoglobin, which is the major tissue chromophore. Therefore, the vasculature appears dark. Also, blue light has a shorter wavelength and hence penetrates superficially compared with red light that has longer wavelength and penetrates deeper, which accentuates the mucosal architecture. These 2 principles combine to enhance superficial mucosal vasculature by optimizing absorbance and scattering of light. Because the vascular density and vascular patterns of adenomatous polyps are different from those of hyperplastic polyps, NBI can help distinguish them in real-time during colonoscopy.
Performance
Initial studies used the Kudo pit pattern to characterize polyp histology with NBI. However, this extrapolation of the Kudo pit pattern to NBI was not believed to be reliable or accurate. In one study, the agreement between the chromoendoscopic and NBI pit patterns was suboptimal, with a kappa of only 0.23. Therefore, newer classifications have emerged, proposed by investigators from different parts of the world. East and colleagues described the vascular pattern intensity classification with NBI. This system refers to the color intensity of the lines surrounding the mucosal pits on the surface of polyps on a 3-point scale. Strong vascular pattern intensity was suggestive of adenomas, and normal or weak pattern intensity was suggestive of hyperplastic polyps ( Fig. 2 , Table 2 ). Based on this classification, 116 polyps less than 10 mm in 62 patients were evaluated in a prospective study with NBI without magnification. The sensitivity, specificity, and accuracy were 94.0%, 89.0%, and 91.4%, respectively, in predicting histology. Similar results were seen for the subgroup analysis of diminutive (≤5 mm) polyps. The authors’ group described a simple classification system with NBI without magnification. Two patterns were described for hyperplastic polyps and 2 for adenomas ( Fig. 3 , Table 3 ). These results showed a sensitivity, specificity, and accuracy of 96%, 89%, and 93%, respectively, for polyp histology prediction. A similar classification was described by Rex that showed an accuracy of 89% for characterizing adenomas. The meshed capillary pattern (CP) classification was proposed by investigators from Japan, using magnification NBI ( Fig. 4 ). CP I was suggestive of hyperplastic polyps, whereas CP II and CP III were diagnostic of adenomas. In a prospective study, the sensitivity, specificity, and accuracy for polyp histology prediction using this classification were 96.4%, 92.3%, and 95.3%, respectively. This Japanese classification was adapted for NBI without magnification in the United States, and showed a diagnostic accuracy of 91% with a sensitivity and negative predictive value of 93% and 91%, respectively. For lesions less than 5 mm, the sensitivity, negative predictive value, and accuracy were 87%, 91%, and 90%, respectively. Another classification based on intensity and shape of small blood vessels on polyps was described from Europe. A fine capillary pattern, with normal size and distribution of vessels, was characteristic of hyperplastic polyps, whereas adenomas showed an increased density, tortuous, corkscrew-type, and branching vascularization ( Fig. 5 ). Using this classification on 200 polyps from 131 patients, NBI with and without magnification showed an accuracy of 91% and 89%, respectively, in differentiating neoplastic from nonneoplastic polyps. Recently, a group of international experts unified various aspects of previously described NBI classifications and proposed and validated the NBI International Colorectal Endoscopic (NICE) classification to predict histology with NBI without magnification ( Fig. 6 , Table 4 ). Using the NICE classification, 118 polyps less than 10 mm were evaluated. The sensitivity, negative predictive value, and accuracy for histology prediction were 98%, 95%, and 89%, respectively.
| Nonneoplastic | Normal | Neoplastic | |
|---|---|---|---|
| Vascular pattern | Weaker (paler) than the surrounding mucosa | The same as the surrounding mucosa | Stronger (darker) than the surrounding mucosa |
| Histology | Surface Mucosal and Vascular Pattern |
|---|---|
| Hyperplastic |
|
| Adenoma |
|
| Type 1 | Type 2 | |
|---|---|---|
| Color | Same or lighter than background | Brown relative to background |
| Vessels | None or isolated lacy vessels | Brown vessels surrounding white structures |
| Surface pattern | Dark or white spots of uniform size or homogeneous absence of pattern | Oval, tubular, or branched white structures surrounded by brown vessels |
| Most likely histology | Hyperplastic polyp | Adenoma |
A meta-analysis on real-time optical diagnosis of polyp histology by NBI included 35 studies and revealed encouraging results in distinguishing neoplastic and nonneoplastic polyps. The reported diagnostic performances were sensitivity, 91.5%; specificity, 85.2%; and negative predictive value, 82.5%. For the 5205 diminutive (<6 mm) polyps that were assessed, the sensitivity was 86.9% and specificity was 84.4%. These results were similar to those of a previous meta-analysis of studies on NBI for differentiating neoplastic from nonneoplastic polyps in real-time that included 28 studies with 6280 polyps. The overall sensitivity was 91.0%, the specificity was 82.6%, and the negative predictive value exceeded 90%. When the diagnosis of diminutive polyps was made with high confidence, then the sensitivity was 93.4% and the specificity was 84.0%.
Feasibility
NBI is a promising imaging technique for polyp histology prediction. It is hassle-free and easy to use, because it can be activated by the push of a button on the handle of the endoscope. Because it is incorporated in the newer and current generation of the Olympus endoscopy system, NBI will be associated with no added costs or extra capital investment once this system is acquired by an endoscopy unit.
Fujinon Intelligent Color Enhancement Technology
FICE uses a postprocessing technique in which the endoscopic image from the processor is altered by computerized spectral estimation technology. The reflected photons of the original image are arithmetically processed to reconstitute a virtual image. The system has 10 preprogrammed FICE settings using an optimal set of wavelengths that can be selected on the keyboard. Each one has a different setting for estimated red, green, and blue wavelength. Image color varies as different wavelengths are applied (ie, shorter wavelengths for surface structures and longer for deeper vessels). As a result, enhancement of the superficial mucosal vascular pattern on the polyp can be achieved, which can enable histology prediction.
Performance
An endoscopic capillary-vessel pattern classification for prediction of polyp histology was described using FICE ( Fig. 7 , Table 5 ). Other studies extrapolated the Kudo pit pattern with chromoendoscopy to FICE, with types I and II suggestive of hyperplastic polyps and types III to V suggestive of adenomas.
Related posts:
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
