If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Faculty of Health and Medical Sciences, University of Adelaide, Adelaide, South Australia, AustraliaDepartment of Gastroenterology and Hepatology, Nagoya University Graduate School of Medicine, Nagoya, Aichi, Japan
Surveillance post–endoscopic resection (ER) currently warrants biopsy samples from the resection site scar in most cases, although clinical practice is variable. A classification with standard criteria for scars has not yet been established. We aimed to create and validate a novel classification for post-ER scars by using specific criteria based on advanced imaging.
Key endoscopic features for scars with and without recurrence were (1) dark brown color, elongated/branched pit pattern, and dense capillary pattern and (2) whitish, pale appearance, round/slightly large pits, and irregular sparse vessels. Scars were first assessed with high-definition white-light endoscopy (HD-WLE) followed by interrogation with narrow-band imaging (NBI). Scars with at least 2 concordant characteristics were diagnosed with “high confidence” for NBI for scar (NBI-SCAR) classification. The final endoscopic predictions were correlated with histopathology. The primary outcome was the difference in sensitivity between NBI-SCAR and HD-WLE predictions. Secondary outcomes included the validation of our findings in 6 different endoscopy settings (Australia, United States, Japan, Brazil, Singapore, and Malaysia). The validation took place in 2 sessions separated by 2 to 3 weeks, each with 10 one-minute videos of post-ER scars on underwater NBI with dual focus. Inter-rater and intrarater reliability were calculated with Fleiss’ free-marginal kappa and Bennett et al. S score, respectively.
One hundred scars from 82 patients were included. Ninety-five scars were accurately predicted with high confidence by NBI-SCAR in the exploratory phase. NBI-SCAR sensitivity was significantly higher compared with HD-WLE (100% vs 73.7%, P < .05). In the validation phase, similar results were found for endoscopists who routinely perform colonoscopies and use NBI (sensitivity of 96.4%). The inter-rater and intrarater reliability throughout all centers were, respectively, substantial (κ = .61) and moderate (average S = .52) for this subset.
NBI-SCAR has a high sensitivity and negative predictive value for excluding recurrence for endoscopists experienced in colonoscopy and NBI. In this setting, this approach may help to accurately evaluate or resect scars and potentially mitigate the burden of unnecessary biopsy samples.
Endoscopic resection (ER) of neoplastic polyps is an important step in colorectal cancer prevention. The timing of follow-up colonoscopy is mainly based on the histologic nature and size of lesions, but some types of resection (eg, piecemeal EMR) are advised to have a closer follow-up for assessment of recurrence.
Follow-up colonoscopy after large piecemeal EMR is usually performed after a 4- to 6-month interval. Larger lesion size, presence of bleeding during the procedure, and high-grade dysplasia have been found to be independent risk factors for recurrence in a multivariate analysis of more than 1000 lesions.
Surveillance colonoscopy post-ER commonly identifies scar tissue at the site of resection, especially after ER of large lesions. Most guidelines recommend that targeted biopsy sampling of the resection site should be performed to exclude histopathologic evidence of recurrence.
Further, sampling error can occur, leading to false-negative results.
Over the years, with development of better endoscopic systems equipped with high-definition (HD) video imaging and magnification narrow-band imaging (NBI), we are now able to examine and describe the mucosal surface in finer detail. The application of advanced endoscopic imaging may increase our capability to rule out recurrence at the post-ER scar site. Several aspects of the scar are used as endoscopic predictors, some of which have been investigated in a recent study by Desomer et al.
Despite describing in detail the steps to evaluate the scar, Desomer et al reverted to the Kudo pit pattern to identify neoplastic pattern. This step, which requires indigo carmine/methylene blue chromoendoscopy, makes the proposal of a simple and direct classification using solely NBI more appealing. NBI also has an advantage over other advanced endoscopic imaging modalities (eg, confocal endomicroscopy), because NBI is much more widely available and easy to use by practicing endoscopists, even outside of expert centers.
demonstrated high accuracy and sensitivity by adopting a systematic approach to look at scars, no formal classification was proposed. This leaves the systematic approach dependent on use of the Kudo classification, which may be more difficult to implement. We propose that the development of a well-defined and simple methodology to evaluate post-ER scars may be an important approach to more accurately identify recurrence.
Therefore, this study was designed to evaluate the use of a simple standardized classification for prediction of recurrence in post-ER scars, which we named NBI for scar (NBI-SCAR) classification.
Because our main objective was to rule out recurrence, a sample size was chosen based on the conservative estimate of 15% sensitivity difference between HD white-light endoscopy (HD-WLE) and NBI-SCAR groups, based on Desomer et al.
For a 95% confidence interval and 80% power, the required number was 97 scars.
Inclusion criteria consisted of patients over 18 years of age referred for colonoscopy at the Endoscopy Unit of Lyell McEwin Hospital who presented a post-ER scar. The colonoscopes used were HQ190 series (Olympus, Tokyo, Japan). Pregnant women, emergency colonoscopies, and patients unwilling to participate were excluded. Standard split-dose bowel preparation with sodium picosulfate and polyethylene glycol was used for all patients. All procedures were performed using carbon dioxide with the patient under sedation by an anesthesiologist or nurse sedationist.
This study was approved by the Human Research Ethics Committee (TQEH/LMH/MH) under the reference number 2008128. This Committee is constituted in accordance with the NHMRC National Statement on Ethical Conduct in Human Research (2007) and incorporating all updates.
All patients had their scars evaluated with a cap and underwater (clear water) technique in regard to 3 features: color, pit pattern, and vascularity. These features were determined by consensus among the authors. The 3 features chosen for the classification were selected mainly based on the well-established colorectal polyp classifications such as the NBI International Colorectal Endoscopic classification.
Hayashi N, Tanaka S, Hewett DG, et al. Endoscopic prediction of deep submucosal invasive carcinoma: validation of the Narrow-Band Imaging International Colorectal Endoscopic (NICE) classification. Gastrointest Endosc 2013;78:625-32.
The exact description of each feature was further refined with an image library of 50 prospectively collected underwater scar images. The conceptualization and determining features for the NBI-SCAR were carried out before the enrollment of the first patient for this study. The 2 endoscopists who participated in the conceptualization of the NBI-SCAR classification also participated in the exploratory phase and have over 10 years (R.S.) and over 3 years (L.Z.C.T.P.) of experience with advanced endoscopic imaging.
Key features of scars with recurrence have been established as follows: dark brown color, elongated or branched pit pattern, and dense capillary pattern surrounding pits. Scars with no recurrence have been established reverse features: a whitish, pale appearance; round and slightly larger pits compared with the surroundings; and irregular sparse vessels with no change in caliber. The NBI-SCAR classification is summarized in Figure 1. If 2 or more concordant key features were found, a high-confidence diagnosis was made. If 1 key feature was found but 2 others were not able to be defined (eg, whitish/pale color, no visible pits, and no discernible vascularity), the scar was predicted with low confidence toward the key feature identified. If none of the 3 key features was identified or if 2 contrasting key features were identified alongside 1 no-identifiable key feature, the diagnosis with the NBI-SCAR classification was deemed as not possible.
The primary outcome was the difference in sensitivity of HD-WLE and NBI-SCAR for detecting recurrence on scars. Secondary outcomes included accuracy measures for overall and high confidence diagnoses with HD-WLE and NBI-SCAR and accuracy and inter/intraobserver agreement of the NBI-SCAR classification in different endoscopy centers around the world (ie, Japan, United States, Brazil, Singapore, Malaysia, and Australia).
In the exploratory phase, the scar was initially evaluated with HD-WLE followed by underwater NBI with dual focus (uNBI-DF) in real time by 2 endoscopists (R.S. and L.Z.C.T.P.). If the 2 endoscopists disagreed regarding 1 feature, that feature was deemed as "not possible to be diagnosed." Scars with at least 2 concordant characteristics were diagnosed with high confidence for NBI-SCAR. HD-WLE diagnosis was considered of high confidence if agreed upon by 2 endoscopists. The final endoscopic prediction was correlated with the biopsy sample of the scar.
The validation phase was conceived in the form of a 2-session test, where both trainees (ie, advanced endoscopy trainees or fellows) and experienced endoscopists (ie, consultants, after formal training) without prior exposure to the NBI-SCAR classification were invited to participate. This was carried out in 6 countries: Japan, United States, Brazil, Singapore, Malaysia, and Australia. None of the invited endoscopist raters was previously familiar with the proposed classification scheme. A formal explanation about the NBI-SCAR was provided in the form of a printed version of the classification, which was given to participants 1 day before the first session took place for familiarization. The recommended time for studying the classification before the test was 10 minutes. Participants were open to ask questions of the site coordinators before or after the test but not while it took place.
The test consisted of 15 short edited videos (<1 minute), which concentrated on the uNBI-DF features of the scar. At first the site coordinator used the first 5 videos to explain the key features they should look for (ie, training videos, 1 with recurrence and 4 without recurrence). The following 10 videos were then presented to the participants (ie, test videos, 1 with and 9 without recurrence), who were required to enter their prediction of each of the 3 key features into a form (Supplementary Figs. 1-3, available online at www.giejournal.org). The videos consisted of cases taken from our initial cohort and had a similar proportion of recurrence within scars as described by the literature on ER. The same test videos, but shuffled, were used 2 to 3 weeks after the initial test for intraobserver reliability evaluation. Only after the second test were the site coordinators authorized to disclose the correct diagnosis for each test video. The set of 20 rated videos per endoscopist was used for accuracy measures calculations.
In addition to responses for each video, the endoscopists were also required to provide information regarding their daily practice and experience. Information was collected on training (under training or finished training), area of interest within endoscopy (eg, luminal endoscopy, EUS), number of procedures performed, frequency of colonoscopy procedures (at least once a week or not), routine use of NBI, and if the endoscopists considered himself or herself an expert in advanced endoscopic imaging. The 3 study phases (conceptualization, exploratory phase, and validation phase) are summarized in Figure 2.
Standard accuracy measures were used to describe the performance of each arm compared with histology. The McNemar test was used to compare HD-WLE and NBI-SCAR prediction in the exploratory phase. Comparison of proportions was performed using the N-1 chi-squared test. Inter-rater reliability was calculated with Fleiss’ free-marginal kappa because there were no restrictions for distribution across categories.
Both inter-rater and intrarater reliability values were interpreted as follows: .0 to .2, slight agreement; .21 to .40, fair agreement; .41 to .60, moderate agreement; .61 to .80, substantial agreement; and .81 to 1.0, almost perfect or perfect agreement. The Wilson score method without continuity correction was used to calculate 95% confidence interval for proportions.
One hundred scars from 82 patients were included in the study. Patient mean age was 67.9 years, and 53% were men. Ninety-five scars enabled diagnosis with high confidence, and 19 had recurrence. The initial histology of the lesions was high-grade dysplastic adenoma in 33% (n = 33), low-grade dysplastic adenoma in 55% (n = 55), dysplastic sessile serrated adenoma/polyp (SSA/P) in 5% (n = 5), and nondysplastic SSA/P in 7% (n = 7). From the 5 scars predicted with low confidence, the index polyp was nondysplastic SSA/P (1) and tubulovillous adenoma (4).
Overall, the NBI-SCAR classification presented 5 false positives and had achieved an overall accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of 95.0%, 100%, 93.8%, 79.2%, and 100%, respectively. HD-WLE presented 5 false negatives and 2 false positives (accuracy = 93.0%, sensitivity = 73.7%, specificity = 97.5%, PPV = 87.5%, NPV = 94.1%). For diagnoses with high confidence, all accuracy measures were raised to 100% with the NBI-SCAR classification (Table 1). NBI-SCAR sensitivity was statistically different from that of HD-WLE (P < .01). The recurrence/residual polyps were all diminutive in size and successfully treated with cold snare/cold avulsion and snare tip soft coagulation.
Table 1Accuracy measures for NBI-SCAR and HD-WLE (exploratory phase)
In the validation phase, across all 6 sites 49 endoscopists were recruited. The baseline characteristics of endoscopists who participated in the validation phase can be found in Table 2. Only 1 center was not an academic center (Malaysia).
Table 2Demographics per center (exploratory phase)
The results varied across centers and was lowest for the nonacademic center (Table 3). The inter-rater reliability for high-confidence diagnoses across centers was moderate (κ = .50; 95% confidence interval, .43-.58), and the average intrarater reliability was fair (average S = .36). Looking at the various subsets, the results for consultants (accuracy = 89.4%, sensitivity = 88.7%, specificity = 89.5%, PPV = 49.1%, NPV = 98.6%) were higher compared to trainee-level endoscopists (accuracy = 84.9%, sensitivity = 76.7%, specificity = 85.8%, PPV = 36.5%, NPV = 97.2%). All comparisons reached statistical significance (p < 0.05) except for specificity (p = 0.09) and NPV (p = 0.13). Optimal results were achieved for endoscopists who perform colonoscopies at least once a week and who are familiar with NBI, which consisted of 24 consultants and 4 trainees (Table 4). The inter-rater reliability for this subgroup across centers was substantial (κ = .61; 95% confidence interval, .51-.71), and the average intrarater reliability was moderate (average S = .52). The inter-rater reliability per center and the intraobserver reliability per endoscopist for high-confidence diagnoses can be found in the Supplementary Tables 1 and 2, respectively (available online at www.giejournal.org).
Table 3Summed high-confidence accuracy measures for NBI-SCAR (validation phase)
Positive predictive value
Negative predictive value
Unites States (n = 12)
Japan (n = 12)
Brazil (n = 13)
Singapore (n = 7)
Malaysia (n = 2)
Australia (n = 3)
Values are % (95% confidence interval).
NBI-SCAR, Narrow-band imaging for scar classification.
The systematic use of biopsy sampling for the evaluation of normal-looking scars has been proposed based on the concern of inconspicuous recurrence that can be missed by the endoscopist. This is illustrated by the study of Knabe et al
conducted from 2010 to 2013 that found a concerning 7% recurrence miss rate for normal-looking scars. However, in the last several years, technology has improved remarkably, and studies have demonstrated the potential for advanced endoscopic imaging to replace the more time-consuming histopathology assessment. For instance, according to Preservation and Incorporation of Valuable Endoscopic Innovations guidelines, advanced imaging could be used to replace histology for polyps with certain characteristics in some scenarios.
(ESCAPE trial) evaluated the use of HD-WLE and NBI with and without magnification for assessment of recurrence within post-ER scars. Similar to the present study, the EMR SCAR group also included an exploratory phase with real-time evaluation of scars and a validation phase that included an “offline” scar assessment by independent endoscopists. There are a few differences in our study compared with the ESCAPE trial
: (1) the suggestion of a standard classification to be used for assessment of scars, (2) the inclusion of not only experts in endoscopic imaging but also trainees and nonexperts, (3) the use of offline videos instead of still images for the validation phase, and (4) the scoring of the same scars after a period of time to assess the intrarater agreement.
One major problem that frequently occurs when aiming to use advanced endoscopic imaging to replace histologic assessment is the lack of consistency among proceduralists because of high interobserver variation.
To address this issue, several classifications have been proposed to standardize the method endoscopists use in evaluating colorectal polyp characteristics. In a not- too-dissimilar manner, this study proposes a standard classification with well-defined criteria to adopt advanced endoscopic imaging as a useful tool in the diagnosis of post-ER scars and recurrence at the scar site. This might facilitate the diagnosis for nonexperts and trainees.
By adopting the NBI-SCAR criteria in the exploratory phase, we were able to achieve an accuracy that is as good as histopathology for excluding neoplastic remnants even with a low-confidence diagnosis. Nevertheless, as with any highly sensitive test, false positives do appear (Fig. 3). However, in this scenario of surveillance colonoscopy post-ER, a false positive is easily dealt with by reverting back to the usual practice of taking samples for histopathology. In contrast, the presence of false negatives would be far more concerning in clinical practice. Fortunately, this was only found in the HD-WLE group (Fig. 4).
One concern that arises when using the NBI-SCAR classification is that it closely relates to the type 2 NBI International Colorectal Endoscopic classification and hence would only be able to predict typical adenoma and not SSA/P recurrence. Although only one of the recurrence cases was because of SSA/P recurrence, this was correctly diagnosed in the exploratory phase. Both recurrence cases included in the validation phase were adenoma recurrences. Further research evaluating the performance of the NBI-SCAR specifically for SSA/P recurrence is warranted.
The principles behind the adoption of uNBI-DF are not only to remove the light reflection cast on by the colonoscope but also to further enhance the mucosal surface pattern. This provides the endoscopist with a magnified and unimpeded view that potentially minimizes the risk of missing neoplastic recurrence. From a more practical standpoint, however, although uNBI-DF appears to offer the best approach in scar site interrogation, this is not always feasible, especially when the area in question is in a difficult location. Thus, occasionally it can be technically challenging to maintain the underwater visualization long enough to properly evaluate the scar and/or for photo documentation. Therefore, although we believe it is useful to image the scar underwater and that uNBI-DF should be used whenever possible for investigating scars, NBI-DF without water immersion is satisfactory for NBI-SCAR as well. We strongly believe that the NBI-SCAR is a useful tool and should be simple enough to use with minimal training involved. Most colonoscopes are equipped with the NBI function, and no further equipment is required to use this technology. Nevertheless, further research is needed to confirm this postulation.
The comparison of imaging modalities in this study was limited to the exploratory phase and based on the dichotomy of HD-WLE and magnified NBI. We did not compare NBI versus NBI-DF. Although there is a possibility that NBI alone would be as accurate as NBI-DF in this study, it is likely that NBI with magnification was the best choice because it was shown to perform better than NBI alone for both polyps
Our study had an unstructured approach rather than a structured methodology (eg, modified Delphi process) to identify which features to use in the classification. However, the NBI-SCAR features were derived from well-established colorectal polyp classifications such as the NBI International Colorectal Endoscopic classification (ie, color, pits, and vascular pattern). Because such features had already been validated as accurate predictors for neoplastic tissue by several studies, a more complicated approach for determining which features to use was not used. In addition, the NBI-SCAR study evaluated the combination of the 3 features but did not interrogate each individual feature for its contributions for the diagnosis of recurrence. Therefore, it is unclear how much each feature would have contributed to the results.
The limitation of our study being conceptualized and evaluated within the same center was overcome with the validation phase. With this we were able to prove that a well-defined classification can lead to consistent results across different centers. In addition, bias from knowing the result from HD-WLE before the use of NBI during the real-time assessment of scars (exploratory phase) was mitigated when similar results were found in the validation phase. The validation of our classification among nonexperts in NBI and gastroenterology trainees suggests that it is easy and simple to learn. However, for optimal results, endoscopists must be familiar with using NBI and perform colonoscopies often. Another possible bias is related to follow-up colonoscopies on the same patients. Eleven of 100 scars included in the exploratory phase were re-evaluated during this period of the study. Although this could bias the endoscopists toward the results from the previous scar assessment, a simple strategy was adopted. During this period of the study, it was advised that for any follow-up colonoscopy, the endoscopists should only assess the endoscopy and pathology reports after they had imaged the scar.
In our center, the use of clips is mainly focused on treating adverse events rather than prophylactic use or to address the open wound per se. In our cohort, only 5 scars had through-the-scope clips used when the initial resection was performed. As highlighted by the ESCAPE and previous trials,
the use of clips might negatively influence accuracy through increased false positives. Therefore, in a center with a routine use of clips post-ER, the performance of the NBI-SCAR classification could be worse. However, the classification focuses on pits, vascularity, and color rather than surface features. Further studies addressing the use of the NBI-SCAR for this specific subset (scars with previous use of clips) are warranted for clarifying this question.
In conclusion, NBI-SCAR has a high sensitivity and NPV for excluding recurrence for endoscopists experienced in colonoscopy and NBI. In this setting, this approach may help to accurately evaluate or resect scars and potentially mitigate the burden of unnecessary biopsy sampling.
We acknowledge Dr Florencia Leiria for her contribution to the exploratory phase data collection.
Supplementary Table 1Inter-rater reliability per center for high-confidence diagnoses
Hayashi N, Tanaka S, Hewett DG, et al. Endoscopic prediction of deep submucosal invasive carcinoma: validation of the Narrow-Band Imaging International Colorectal Endoscopic (NICE) classification. Gastrointest Endosc 2013;78:625-32.
DISCLOSURE: The following authors disclosed financial relationships relevant to this publication: T. M Berzin: Consultant for Fujifilm and Wision AI. All other authors disclosed no financial relationships relevant to this publication.
If you would like to chat with an author of this article, you may contact Dr Singh at [email protected]