Not Ready for Prime Time: Real-Time Narrow-Band Imaging for Small Colorectal Polyps
There is a 5% lifetime risk for developing colorectal cancer, which is the fourth most diagnosed cancer and remains the second leading cause of cancer-related deaths in the United States. Proper screening can prevent colorectal cancer through detection and removal of premalignant neoplastic polyps (adenomas). Colonoscopy is the preferred method of screening, as it facilitates both detection and removal of lesions, although other imaging technologies can be used for polyp detection. While general recommendations are that average-risk individuals be screened at age 50 years, the interval at which follow-up colonoscopy is needed is generally determined based on the findings at this index assessment. Histopathologic analysis is the gold standard for diagnostic analysis of polyps identified during this procedure and is the driving factor in determination of surveillance interval for patients with polyps.
Narrow band imaging (NBI) has become increasingly available over the last decade, as have several other imaging technologies that can be used to improve diagnosis during colonoscopy. In many cases, NBI scopes have both standard and near-focus settings. The latter enables detailed assessment of surface textures and shapes at near microscopic magnifications. This has made possible the development of endoscopic criteria based on NBI that aid in diagnosis based on macroscopic features that correlate with histopathology. However, it is important to note that only surface features can be assessed by NBI and that the magnification provided does not allow imaging of individual cells or detection of cytologic dysplasia.
The investigators of the VALID randomized clinical trial asked whether near-focus NBI diagnosis of small colon polyps was sufficiently sensitive and specific to make histologic diagnosis unnecessary. The investigators studied 558 patients undergoing screening colonoscopy who were randomized to groups assessed with either near-focus or standard-view optical colonoscopes. Both technologies allowed endoscopists to make diagnoses with high confidence in the majority of cases. Near-focus NBI provided statistically significant improvement in the rate of high-confidence diagnoses relative to standard-view optical diagnosis (85.1% vs 72.6%) when assessing polyps <5 mm in diameter. Regardless of the imaging modality, slightly over 90% of high-confidence diagnoses correlated with histopathologic diagnoses. The negative and positive predictive values of the high-confidence diagnoses were also over 90%. In contrast, the accuracy, sensitivity, specificity, negative predictive value, and positive predictive value were all reduced substantially in the low-confidence diagnoses, and, in these, near-focus NBI actually performed worse than standard-view NBI. Although the data were not reported, a rough calculation from the information provided suggests that the overall diagnostic accuracy using near-focus NBI was only approximately 80%. Nevertheless, the authors propose that small polyps can be adequately diagnosed using near-focus NBI and then discarded without histopathologic analysis. This is in agreement with a recent proposal by the American Society for Gastrointestinal Endoscopy (ASGE), and raises several concerns.
First, it is important to recognize the limitations of these technologies. As noted above, the value drops substantially when features allowing high-confidence diagnoses are not present. Further, NBI approaches evaluate only surface features and cannot distinguish sessile serrated adenomas/polyps (SSA/Ps) from traditional serrated adenomas or hyperplastic polyps. The populations studied had an unusually low incidence of SSA/Ps, which may have artificially amplified apparent diagnostic accuracy. This may explain why this study differs from several recent reports that failed to demonstrate this degree of diagnostic accuracy. In addition, it is important to note that SSA/Ps are thought to be precursor lesions of MSI-high colorectal cancers, which are associated with hereditary non–polyposis colorectal cancer (HNPCC, or Lynch syndrome). Thus, regardless of accuracy, the NBI approach is unlikely to be adequate in assessment of patients with a family history suspicious for HNPCC.
Two other primary benefits of dispensing with histopathologic diagnoses were noted. First, if the final diagnosis is made endoscopically, patients can be given final diagnoses at discharge from the endoscopy suite. This saves the up to 2-day delay between polyp biopsy and availability of a final pathologic diagnosis. However, given that these results are used to determine the need for repeat colonoscopy at intervals from 2 to 10 years, making the diagnosis at time of colonoscopy cannot provide any benefit in terms of clinical care.
A second argument in favor of discarding polyps following endoscopic diagnosis is that this would save vast sums currently expended in pursuit of histopathologic diagnoses. However, when examined closely, the logic supporting this approach falls away. Current Medicare reimbursement rates for colonoscopy with polyp biopsy or resection fall between $700 and $900. On the basis of current coding criteria, these can be doubled if two polyps are present and are removed using different procedures. Compare this to the approximately $70 Medicare reimbursement for each biopsy container evaluated microscopically by a surgical pathologist. This cost includes both the technical and professional components of the fees, and, by current policies, is the same whether the biopsy container includes 1 or 10 polyps. Thus, histopathologic diagnosis has a minimal impact on overall cost of colonoscopy and polyp removal. It is not unreasonable to group small polyps in single containers if the goal is to reduce pathology costs. As long as these polyps are small and extremely unlikely to harbor malignant lesions where identification of the site of the polyp is important to future therapy, there is no harm to the patient, as surveillance interval is based on the entire population rather than individual polyps.
Importantly, all of the discussion above relates only to small polyps. Because larger polyps carry a risk of malignancy that is roughly proportional to polyp size, it would be appropriate to put larger polyps in individual containers. This would be critical if, for example, an invasive lesion requiring segmental colectomy was identified. Further, even for small polyps diagnosed with high confidence, the present study had a 3.6% false-negative rate. This means that up to 3.6% of patients thought to be free of adenomas were not. While not all adenomas progress to invasive cancer, available data do show that it is important to shorten the surveillance interval in patients with adenomas. In this respect, a resect-and-discard strategy could lead to an inappropriately lengthened surveillance interval and subsequent development of colorectal cancer in a small portion of patients. To endorse these strategies, we must be willing to accept the significant ramifications for this subset of patients and their families, which includes an approximately $30,000 increase in cancer-related healthcare costs per patient (in 2005 dollars) as well as increased liability for endoscopists.
Despite our concerns that failure to perform histopathologic diagnosis of resected polyps introduces unnecessary risk to our patients, we believe advanced imaging technologies, including near-focus NBI, are exciting developments with significant utility. For example, in inflammatory bowel disease patients with numerous polyps, many which are suspected to be inflammatory pseudo-polyps, this in situ evaluation tool might allow targeted resection of the most worrisome lesions when it is impossible to resect all polyps. In addition, we must recognize that histopathologic diagnosis requires that the lesion be present on the slide examined. In some cases, this does not occur, either because the lesion was inadequately sampled in the endoscopic biopsy or because the sections, although prepared in the histopathology laboratory, failed to demonstrate diagnostic regions. Risk of the latter cause of error is reduced by the standard practice of examining several levels of sections through each biopsy, but such errors do still occur. Risk of misdiagnosis could be further reduced in some cases if, for example, further sections are cut when an endoscopic high-confidence diagnosis of adenoma is not confirmed histopathologically. This is similar to current practice in evaluating cervical pathology, where the standard of care requires that cytology and biopsy results be correlated. Because NBI diagnosis only added a few seconds to the time required for identification and resection of polyps, providing this information could be of great benefit at little or no additional cost. Finally, observer error is possible in surgical pathology as in any subjective interpretation. This will depend, in part, on the skill level of the observer. For example, is the surgical pathologist subspecialty-trained in gastrointestinal pathology? A similar concern can be raised for endoscopic interpretation, as other studies in both academic and non-academic settings have failed to achieve the high levels of accuracy achieved by the expert endoscopists involved in the VALID trial.
Overall, advances in endoscopic imaging are exciting and hold great promise. However, at present, it is far too soon to discard the gold standard of histopathologic diagnosis.
Click on any of these tags to subscribe to Topic Alerts. Once subscribed, you can get a single, daily email any time PracticeUpdate publishes content on the topics that interest you.
Visit your Preferences and Settings section to Manage All Topic Alerts
Additional Info
Disclosure statements are available on the authors' profiles: