Chromoendoscopy and Narrow-Band Imaging Compared With High-Resolution Magnification Endoscopy in Barrett’s Esophagus
2008; Elsevier BV; Volume: 134; Issue: 3 Linguagem: Inglês
10.1053/j.gastro.2008.01.003
ISSN1528-0012
AutoresWouter L. Curvers, Lubbertus C. Baak, Ralf Kießlich, Arnoud van Oijen, Thomas Rabenstein, Krish Ragunath, Jean‐François Rey, Pieter Scholten, Uwe Seitz, Fiebo ten Kate, Paul Fockens, Jacques Bergman,
Tópico(s)Esophageal and GI Pathology
ResumoBackground & Aims: The aim of this study was to compare magnified still images obtained with high-resolution white light endoscopy, indigo carmine chromoendoscopy, acetic acid chromoendoscopy, and narrow-band imaging to determine the best technique for use in Barrett’s esophagus. Methods: We obtained magnified images from 22 areas with the 4 aforementioned techniques. Seven endoscopists with no specific expertise in Barrett’s esophagus or advanced imaging techniques and 5 international experts in this field evaluated these 22 areas for overall image quality, mucosal image quality, and vascular image quality. In addition, the regularity of mucosal and vascular patterns and the presence of abnormal blood vessels were evaluated, and this was correlated with histology. Results: The interobserver agreement for the 3 features of mucosal morphology with white light images ranged from κ = 0.51 (95% confidence interval [CI]: 0.46–0.55) to κ = 0.53 (95% CI: 0.50–0.57) for all observers, from κ = 0.43 (95% CI: 0.33–0.54) to κ = 0.53 (95% CI: 0.41–0.64) for experts, and from κ = 0.51 (95% CI: 0.15–0.33) to κ = 0.64 (95% CI: 0.58–0.70) for nonexperts. The interobserver agreement in these groups did not improve by adding one of the enhancement techniques. The yield for identifying early neoplasia with white light images was 86% for all observers, 90% for experts, and 84% for nonexperts. The addition of enhancement techniques did not improve the yield neoplasia. Conclusions: The addition of indigo carmine chromoendoscopy, acetic acid chromoendoscopy, or narrow-band imaging to white light images did not improve interobserver agreement or yield identifying early neoplasia in Barrett’s esophagus. Background & Aims: The aim of this study was to compare magnified still images obtained with high-resolution white light endoscopy, indigo carmine chromoendoscopy, acetic acid chromoendoscopy, and narrow-band imaging to determine the best technique for use in Barrett’s esophagus. Methods: We obtained magnified images from 22 areas with the 4 aforementioned techniques. Seven endoscopists with no specific expertise in Barrett’s esophagus or advanced imaging techniques and 5 international experts in this field evaluated these 22 areas for overall image quality, mucosal image quality, and vascular image quality. In addition, the regularity of mucosal and vascular patterns and the presence of abnormal blood vessels were evaluated, and this was correlated with histology. Results: The interobserver agreement for the 3 features of mucosal morphology with white light images ranged from κ = 0.51 (95% confidence interval [CI]: 0.46–0.55) to κ = 0.53 (95% CI: 0.50–0.57) for all observers, from κ = 0.43 (95% CI: 0.33–0.54) to κ = 0.53 (95% CI: 0.41–0.64) for experts, and from κ = 0.51 (95% CI: 0.15–0.33) to κ = 0.64 (95% CI: 0.58–0.70) for nonexperts. The interobserver agreement in these groups did not improve by adding one of the enhancement techniques. The yield for identifying early neoplasia with white light images was 86% for all observers, 90% for experts, and 84% for nonexperts. The addition of enhancement techniques did not improve the yield neoplasia. Conclusions: The addition of indigo carmine chromoendoscopy, acetic acid chromoendoscopy, or narrow-band imaging to white light images did not improve interobserver agreement or yield identifying early neoplasia in Barrett’s esophagus. See Yachimski P et al on page 302 in CGH; See CME quiz on page 866. Chromoendoscopy is a technique in which staining agents are sprayed on the mucosal surface of the gastrointestinal tract to enhance the endoscopic evaluation of the mucosa. This enables endoscopic detection of surface patterns or functional characteristics that may predict the presence of early neoplasia, lesions that may be difficult to detect with standard endoscopy only. Chromoendoscopy is usually combined with magnifying endoscopy for best results. In Barrett’s esophagus (BE), the typically used contrast enhancement agents are indigo carmine and acetic acid. Indigo carmine enhances the mucosal surface by pooling in the grooves between the mucosal villi, thus enabling the visualization of the pattern formed by mucosal folds and pits. Acetic acid achieves the same goal by means of reversible denaturation of superficial mucosal proteins. Both agents have been reported to recognize typical arrangements of the mucosal pit patterns that may correlate with histology.1Sharma P. Weston A.P. Topalovski M. et al.Magnification chromoendoscopy for the detection of intestinal metaplasia and dysplasia in Barrett’s oesophagus.Gut. 2003; 52: 24-27Google Scholar, 2Guelrud M. Herrera I. Essenfeld H. et al.Enhanced magnification endoscopy: a new technique to identify specialized intestinal metaplasia in Barrett’s esophagus.Gastrointest Endosc. 2001; 53: 559-565Google Scholar Recently, a new optical technique called narrow-band imaging (NBI) has been developed that may replace chromoendoscopy for imaging of the gastrointestinal tract. NBI uses the optical phenomenon that the depth of light propagation into tissues depends on its wavelength. Light of a short wavelength (ie, blue light in the visible spectrum) penetrates only superficially into the mucosa, allowing for improved surface detail imaging. In NBI, simple optical filters are used to achieve the necessary contrast-enhancement effect. In addition, NBI has the advantage of revealing the vascular patterns with a high contrast because the used blue light is highly absorbed by hemoglobin.3Gono K. Obi T. Yamaguchi M. et al.Appearance of enhanced tissue features in narrow-band endoscopic imaging.J Biomed Opt. 2004; 9: 568-577Google Scholar We evaluated the features of dysplastic and nondysplastic BE in a standardized manner and found that nondysplastic BE was associated with regular mucosal and vascular patterns without abnormal blood vessels and that dysplastic BE was associated with irregular mucosal and/or vascular patterns and/or the presence of abnormal blood vessels.4Kara M.A. Ennahachi M. Fockens P. et al.Detection and classification of the mucosal and vascular patterns (mucosal morphology) in Barrett’s esophagus by using narrow band imaging.Gastrointest Endosc. 2006; 64: 155-166Google Scholar The aim of this study was to compare images obtained with magnifying high-resolution white light endoscopy (WLE), NBI, indigo carmine chromoendoscopy (ICC), and acetic acid chromoendoscopy (AAC) for imaging quality and evaluation of mucosal morphology to determine the best technique for use in BE. Patients were eligible for this study when scheduled for either a surveillance endoscopy of a known BE or referred for workup of high-grade intraepithelial neoplasia (HGIN) or early cancer (EC). Patients with advanced esophageal cancers were excluded as were patients with BE with a length smaller than 3 cm. The protocol was approved by the Medical Ethics Committee of our institution, and informed consent was obtained from all patients prior to the procedure. All procedures were performed using a high-resolution magnification endoscope (GIF Q240Z/GIF-Q240FZ; Olympus Inc, Tokyo, Japan). A transparent distal attachment (D-201-11802; Olympus Inc) with a diameter of 11.8 mm and a free distal distance of 2 mm was attached to the endoscope to fix mucosal areas of interest until images were taken. The endoscopic imaging system consisted of the Olympus XCV-260HP processor and Olympus CV XCLV-260HP light source (Olympus Inc). This red, green, blue (RGB) sequential illumination light source contains 2 separate sets of RGB filters, which can be used interchangeably by pressing a button located on the light source. Each set of RGB filters is situated in a rotating disk, which is synchronized with a monochromatic charge-coupled device for image detection. The 3 detected images are integrated by the video processor into the final color image, which is displayed on a standard video monitor. One of the rotating disks contains conventional (broad band) RGB filters, which are used for WLE. The second disk contains special filters designed for NBI. Digital still images taken during endoscopy were automatically saved on a connected computer. The NBI endoscopy system was provided by Olympus Inc, Tokyo, Japan. All procedures were performed by the same endoscopist (J.B.), who used intravenous midazolam (2.5 to 15 mg) to place patients under conscious sedation. The esophagus was first inspected for the presence of a columnar-lined segment and a hiatal hernia. The lengths of the BE segment and hiatal hernia were recorded according to the Prague C & M classification.5Sharma P. Dent J. Armstrong D. et al.The development and validation of an endoscopic grading system for Barrett’s esophagus: the Prague C & M criteria.Gastroenterology. 2006; 131: 1392-1399Abstract Full Text Full Text PDF Scopus (799) Google Scholar Visible abnormalities were defined according to the macroscopic classification of early neoplasia of the digestive tract.6Endoscopic Classification Review GroupUpdate on the Paris classification of superficial neoplastic lesions in the digestive tract.Endoscopy. 2005; 37: 570-578Google Scholar, 7The Paris endoscopic classification of superficial neoplastic lesions: esophagus, stomach, and colon: November 30 to December 1, 2002.Gastrointest Endosc. 2003; 58: S3-S4Google Scholar Subsequently, 10–20 cc of acetylcysteine was applied using a spraying catheter (PW-SW-1; Olympus Inc) passed through the accessory channel of the endoscope to dissolve any mucus that might reduce the quality of imaging. One to 2 minutes after application of the acetylcysteine, the esophagus was washed with 50–100 cc tap water to remove the dissolved mucinous layer. Next, a number of randomly selected areas (upon the discretion of the endoscopist) were stabilized using the distal attachment cap and magnified for visualization of the mucosal morphology using WLE, NBI, ICC (0.5% indigo carmine), and AAC (2% acetic acid), respectively. Images were obtained first with WLE using the zoom function of the endoscope in such a way that the rim of the distal attachment cap was just visible or had just left the field of view. Subsequently, the NBI filters were activated and images were obtained without changing the position of the endoscope or the zoom mode. Indigo carmine (0.5%; 5–10 cc) was then applied through the spraying catheter on the same area of interest followed by repositioning the endoscope and obtaining still images. Subsequently, the indigo carmine was rinsed off using careful water flushes until all blue staining had disappeared from the BE segment followed by application of 5–10 cc of acetic acid (2%) through the spraying catheter. Again, the endoscope was repositioned onto the same area, and still images were obtained. Areas were selected in such a way that small landmarks (eg, a small island of squamous mucosa) enabled repositioning of the endoscope as accurate as possible onto the same area during the sequence of imaging using feedback from the images obtained with the other imaging techniques as a reference. Of all areas that were imaged, 3 to 6 still images were taken with each technique and stored as Bitmap (BMP) files. After all images had been obtained, at least 2 targeted biopsy specimens were obtained from each area. All biopsy specimens were routinely processed and evaluated by 2 pathologists (a junior staff member supervised by a senior pathologist). For the purpose of this study, all biopsy specimens were revised by one gastrointestinal pathologist (F.T.K.) with extensive experience in BE8Dixon M.F. Gastrointestinal epithelial neoplasia: Vienna revisited.Gut. 2002; 51: 130-131Google Scholar and classified according to the revised Vienna classification for gastrointestinal neoplasia. From all areas, the best image obtained with each technique was selected based on overall imaging quality and similarity among the images taken with the different techniques from one area. None of these images underwent any form of postprocessing after being obtained during the endoscopic procedure. All images were incorporated in a slide show (Microsoft Office PowerPoint 2003; Microsoft Inc, Redmond, WA) and had the same size and resolution. The images were evaluated by 7 Dutch endoscopists from 6 nonuniversity hospitals with no specific expertise in BE or advanced endoscopic imaging techniques evaluated in this study. In addition, 5 European endoscopists with experience in advanced imaging techniques and in BE also evaluated these images (R.K., T.R., K.R., J–F.R., U.S.). All expert observers are well-known in the international field for the endoscopic detection and treatment of early neoplasia in BE. They have published on this subject, and the techniques described in this study are all available at their centers.9Kiesslich R. Hahn M. Herrmann G. et al.Screening for specialized columnar epithelium with methylene blue: chromoendoscopy in patients with Barrett’s esophagus and a normal control group.Gastrointest Endosc. 2001; 53: 47-52Google Scholar, 10Kiesslich R. Neurath M.F. Galle P.R. Chromoendoscopy and magnifying endoscopy in patients with gastroesophageal reflux disease Useful or negligible?.Dig Dis. 2004; 22: 142-147Google Scholar, 11Pech O. Gossner L. May A. et al.Long-term results of photodynamic therapy with 5-aminolevulinic acid for superficial Barrett’s cancer and high-grade intraepithelial neoplasia.Gastrointest Endosc. 2005; 62: 24-30Google Scholar, 12Pohl J. May A. Rabenstein T. et al.Comparison of computed virtual chromoendoscopy and conventional chromoendoscopy with acetic acid for detection of neoplasia in Barrett’s esophagus.Endoscopy. 2007; 39: 594-598Google Scholar, 13Anagnostopoulos G.K. Yao K. Kaye P. et al.Magnification endoscopy with narrow band imaging in Barrett’s esophagus.J Clin Gastroenterol. 2006; 40: S192-S193Google Scholar, 14Fortun P.J. Anagnostopoulos G.K. Kaye P. et al.Acetic acid-enhanced magnification endoscopy in the diagnosis of specialized intestinal metaplasia, dysplasia and early cancer in Barrett’s oesophagus.Aliment Pharmacol Ther. 2006; 23: 735-742Google Scholar, 15Ragunath K. Krasner N. Raman V.S. et al.A randomized, prospective cross-over trial comparing methylene blue-directed biopsy and conventional random biopsy for detecting intestinal metaplasia and dysplasia in Barrett’s esophagus.Endoscopy. 2003; 35: 998-1003Google Scholar, 16Rey J.F. Inoue H. Guelrud M. Magnification endoscopy with acetic acid for Barrett’s esophagus.Endoscopy. 2005; 37: 583-586Google Scholar, 17Rey J.F. Kuznetsov K. Sattonnet C. et al.Diagnosis of Barrett’s esophagus with narrow band imaging endoscopy.Gastrointest Endoscopy. 2004; 59: P92Google Scholar, 18Seitz U. Soehendra N. Endoscopy: current state and future trends in tumor diagnosis.Anticancer Res. 2003; 23: 827-829Google Scholar, 19Seewald S. Akaraviputh T. Seitz U. et al.Circumferential EMR and complete removal of Barrett’s epithelium: a new approach to management of Barrett’s esophagus containing high-grade intraepithelial neoplasia and intramucosal carcinoma.Gastrointest Endosc. 2003; 57: 854-859Google Scholar All observers were informed that the purpose of the study was to compare different techniques for detailed imaging of areas in BE, but they were left unaware of the technical background of the imaging techniques used. Prior to evaluation of the images, the observers were given a 10-minute automated presentation on endoscopic mucosal morphology in BE. In this presentation, the different characteristics of mucosal morphology in BE were explained based on our recent results on mucosal morphology in BE.4Kara M.A. Ennahachi M. Fockens P. et al.Detection and classification of the mucosal and vascular patterns (mucosal morphology) in Barrett’s esophagus by using narrow band imaging.Gastrointest Endosc. 2006; 64: 155-166Google Scholar Eight examples of regular mucosal and vascular patterns, 4 examples of flat mucosa with long branching vessels, and 7 examples of irregular mucosal and vascular patterns with abnormal blood vessels were shown. To avoid bias, these examples were shown with WLE images only. Subsequently, the different scoring forms were explained. Before starting the study evaluation, 10 WLE images (6 nondysplastic BE and 4 HGIN/EC) were assessed as a learning set, with the observers receiving feedback from the principal investigator supervising the assessment immediately after scoring each image. The images used for the demonstration and the learning set were not included in the final evaluation. The sequence of images in the slide show was presented as follows: WLE images of all areas were shown in random order with areas of different patients being mixed. This assessment was performed to measure the appreciation of the image quality of the WLE images without bias because of the observation of images obtained with the other techniques. From each image, the observers evaluated 3 image quality characteristics: overall image quality, mucosal folds and/or mucosal pits image quality, and image quality of blood vessels on a 10-cm visual analogue scale (VAS). Subsequently, the observers evaluated the mucosal morphology of the area by answering the following questions: (1) Is the mucosal pattern on this slide regular, irregular, or flat; (2) is the vascular pattern on this slide regular or irregular; (3) are there abnormal blood vessels present? After all WLE images had been evaluated separately, WLE images were shown combined with an image from the same area obtained with either NBI, ICC, or ACC. Again, these slides were shown in random order with different patients, different areas, and different techniques being mixed. This assessment was performed to measure the appreciated characteristics of image quality of the NBI, ICC, and ACC images compared with the corresponding WLE images in an independent manner without bias because of the combined observation of all images obtained with the different techniques. During this second evaluation, the 3 characteristics of image quality were scored on an ordinal scale. This scale ranged from −3 (much worse than WLE), −2 (worse than WLE), −1 (little worse than WLE), 0 (same as WLE), +1 (little better than WLE), +2 (better than WLE) to +3 (much better than WLE). Subsequently, the mucosal morphology was evaluated as aforementioned, using the information from both images. Finally, all 4 corresponding images of each area imaged were presented together (Figure 1). To avoid bias, the position of the NBI, ICC, and ACC images were changed randomly between the different slides. Again, the slides were shown in a random order with patients and areas being mixed. The observers ranked the 4 images from best (rank 1) to worst (rank 4) for the 3 characteristics of image quality. Finally, the mucosal morphology was evaluated with the information available from all 4 techniques. All calculations were carried out using the Statistical Package for Social Sciences for Windows 12.0.1 software package (SPSS Inc, Chicago, IL). Kappa statistics with their 95% confidence intervals (CI) were used to test for interobserver agreement of the 3-step classification system using arbitrary interpretation by Landis and Koch (0, poor agreement; 0.00–0.20, slight agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, substantial agreement; 0.80–1.00, almost perfect agreement).20Landis J.R. Koch G.G. The measurement of observer agreement for categorical data.Biometrics. 1977; 33: 159-174Google Scholar Because κ statistics can only be calculated with pair-wise observations, κ values were calculated for all pair-wise combinations of the observers. To determine the yield of the observers for identifying HGIN/EC, an area was called suspicious when at least 1 of the 3 following features was present according to the observers: irregular mucosal pattern, irregular vascular pattern, or the presence of abnormal blood vessels.4Kara M.A. Ennahachi M. Fockens P. et al.Detection and classification of the mucosal and vascular patterns (mucosal morphology) in Barrett’s esophagus by using narrow band imaging.Gastrointest Endosc. 2006; 64: 155-166Google Scholar We defined the yield of HGIN/EC of each technique as the number of assessments of areas with HGIN/EC that were scored positive for at least 1 of the 3 aforementioned features divided by the total number of assessments of areas with HGIN/EC. In total, 14 patients (11 males; mean age, 73 years [SD, 10]) were included in this study: 9 were referred for work-up for treatment of HGIN/EC, 3 underwent follow-up after endoscopic treatment, and 2 underwent standard BE surveillance. The median circumferential extent of the BE segment was 5 cm (interquartile range [IQR], 5–7), and the median maximal extent was 6 cm (IQR, 6–10). In total, 28 areas were imaged with all 4 techniques. Six areas were excluded because of insufficient imaging quality or not enough similarity between the areas imaged with all 4 techniques. In total, 22 areas were included in the evaluation. The histologic diagnoses of the areas were as follows: 8 areas with HGIN/EC, 1 area with low-grade dysplasia, 1 area indefinite for dysplasia, and 12 areas with nondysplastic BE. WLE images were in general considered to be of good quality: the median VAS scores for overall image quality, mucosal imaging quality, and vascular imaging quality were 7.5 (IQR, 6.0–8.5), 7.2 (IQR, 6.0–8.4), and 7.1 (IQR, 6.0–8.4), respectively. In the pair-wise comparisons between WLE and either NBI, ICC, or AAC, differences were expressed on an ordinal scale ranging from −3 (much worse than WLE), −2 (worse than WLE), −1 (little worse than WLE), 0 (same as WLE), +1 (little better than WLE), +2 (better than WLE) to +3 (much better than WLE). For analysis, the evaluations were divided in 3 groups: (1) better than WLE (+1, +2, and +3); (2) same as WLE (0); and (3) worse than WLE (−1, −2, and −3) (Table 1).Table 1Appreciated Characteristics of Image Quality of Magnified Still Images of Barrett’s Esophagus Obtained With Narrow Band Imaging (NBI), Indigo Carmine Chromoendoscopy (ICC), or Acetic Acid Chromoendoscopy (AAC) Compared With High-Resolution White Light Endoscopy (WLE) Images of the Corresponding AreaAppreciated characteristics of image qualities compared to WLE (%)Overall image qualityMucosal image qualityVascular image qualityBetterEqualWorseBetterEqualWorseBetterEqualWorseAll observers NBI762136334368284 ICC513415642312244234 AAC73171080128363331Experts NBI90825640481154 ICC622414662410174637 AAC7811118686163252Non-experts NBI672946830259374 ICC424216622216294031 AAC69229751510493318 Open table in a new tab When all observers were grouped together, all 3 enhancement techniques (NBI, ICC, or AAC) scored better than WLE for almost each image quality characteristic, except ICC for vascular image quality. NBI and AAC were best appreciated for overall image quality: 76% and 73%, respectively, of pair-wise comparisons with WLE were scored as “better than WLE.” AAC was best appreciated for mucosal image quality: in 80% of pair-wise evaluations between AAC and WLE, AAC mucosal image quality was scored as “better than WLE.” NBI was best appreciated for vascular image quality: in 60% of pair-wise evaluations between NBI and WLE, the former was scored as “better than WLE.” Compared with nonexperts, the appreciation of the different image quality characteristics by experts showed a more explicit difference in the pair-wise comparisons of WLE and the 3 enhancement techniques. For overall image quality, NBI and ICC were more frequently scored as “better than WLE” in pair-wise comparisons with WLE by experts than by nonexperts (NBI: 90% vs 67%, respectively, P < .01; ICC: 62% vs 42%, respectively, P < .01). For vascular image quality, NBI was more frequently scored as “better than WLE” by experts than by nonexperts (81% vs 59%, respectively, P < .01). In addition, experts scored AAC more often “worse than WLE” than nonexperts for vascular imaging quality (52% vs 18%, respectively, P < .01). The observers ranked the 4 images (WLE, NBI, ICC, and AAC) of all 22 areas from 1 (best technique) to 4 (worst technique) (Table 2). When all observers were grouped together, NBI and AAC were ranked most frequently as number 1 (best technique) for overall image quality (43% and 40%, respectively, of the comparisons), and WLE and ICC were ranked most often as number 4 (worst technique) (38% and 39%, respectively, of the comparisons). For mucosal image quality, the majority ranked AAC as number 1 (56%) and WLE as number 4 (49%). The majority of observers ranked NBI as number 1 for vascular image quality (69%), whereas AAC was ranked as number 4 by the majority (52%).Table 2Evaluation by Expert and Non-Expert Endoscopists of Four Corresponding Magnified Still Images From Barrett’s Esophagus Obtained With Either High-Resolution White Light Endoscopy (WLE), Narrow Band Imaging (NBI), Indigo Carmine Chromoendoscopy (ICC), or Acetic Acid Chromoendoscopy (AAC)All observersRankOverall image quality (%)Mucosal image quality (%)Vascular image quality (%)123412341234HRE522313821929496483212NBI4334165282833106917104ICC82029391332252811243430AAC402021165619111213102252ExpertsHRE1103848262960145474NBI573291262737789800ICC10283027204017214393123AAC292719205124149561668Non-expertsHRE8322531227294010512118NBI3335217292928125423167ICC7142947827313316133634AAC47152112591691420132740NOTE. For each item the observer ranked, the 4 techniques from “1” (best of the 4, shown in bold) to “4” (worst of the 4, shown in shaded bars). Open table in a new tab NOTE. For each item the observer ranked, the 4 techniques from “1” (best of the 4, shown in bold) to “4” (worst of the 4, shown in shaded bars). For overall imaging quality, nonexperts ranked AAC (47%) most frequently as number 1, whereas experts ranked NBI (57%) the most often as number 1. The majority of the experts as well as nonexperts ranked NBI as number 1 for vascular image quality. The proportion of experts, however, that ranked NBI as number 1 for this item was higher than the proportion of nonexperts (89% vs 54%). A larger proportion of experts ranked AAC as number 4 compared with nonexperts (68% vs 40%, respectively). When the evaluations of all observers were grouped together, the interobserver agreement on 3 items of mucosal morphology assessed on WLE images was moderate (mucosal pattern: κ = 0.53 [95% confidence interval (CI): 0.50–0.57], vascular pattern: κ = 0.51 [95% CI: 0.46–0.55], presence of abnormal blood vessels: κ = 0.52 [95% CI: 0.48–0.57]). When WLE was combined with NBI, ICC, or AAC, the interobserver agreement did not improve, and the same was noted when all techniques were evaluated together (Table 3). The κ values did not substantially differ after exclusion of the 2 areas that contained low-grade dysplasia and indefinite for dysplasia from the analysis.Table 3Inter-Observer Agreement Amongst Expert and Non-expert Endoscopists in Evaluating (ir)Regularity of Mucosal Patterns, (ir)Regularity of Vascular Patterns, and the Presence of Abnormal Blood Vessels of Magnified Still Images of Barrett’s Esophagus Obtained With Either High-Resolution White Light Endoscopy (WLE), or Combinations of WLE With Narrow Band Imaging (NBI), Indigo Carmine Chromoendoscopy (ICC) or Acetic Acid Chromoendoscopy (AAC), and all 4 Techniques TogetherWLE (k)WLE-NBI (k)WLE-ICC (k)WLE-AAC (k)All (k)All observers Mucosal pattern0.53 (0.50–0.57)0.39 (0.35–0.43)0.46 (0.42–0.50)0.42 (0.38–0.46)0.43 (0.39–0.48) Vascular pattern0.51 (0.46–0.55)0.42 (0.37–0.46)0.44 (0.39–0.49)0.48 (0.43–0.52)0.45 (0.41–0.50) Presence of abnormal blood vessels0.52 (0.48–0.57)0.40 (0.35–0.44)0.32 (0.27–0.37)0.42 (0.37–0.47)0.41 (0.36–0.46) Yield of identifying HGIN/EC (%)8684708384Experts Mucosal pattern0.43 (0.33–0.53)0.18 (0.07–0.28)0.46 (0.34–0.57)0.35 (0.25–0.45)0.34 (0.22–0.45) Vascular pattern0.52 (0.40–0.63)0.45 (0.33–0.57)0.45 (0.33–0.58)0.44 (0.32–0.55)0.44 (0.32–0.56) Presence of abnormal blood vessels0.53 (0.41–0.64)0.45 (0.33–0.57)0.42 (0.29–0.55)0.53 (0.41–0.65)0.61 (0.49–0.73) Yield of identifying HGIN/EC (%)9085688083Non-experts Mucosal pattern0.64 (0.58–0.70)0.51 (0.45–0.58)0.47 (0.40–0.53)0.51 (0.44–0.58)0.51 (0.44–0.58) Vascular pattern0.51 (0.44–0.59)0.38 (0.30–0.47)0.41 (0.33–0.49)0.50 (0.42–0.57)0.43 (0.35–0.51) Presence of abnormal blood vessels0.51 (0.43–0.59)0.36 (0.28–0.45)0.24 (0.15–0.33)0.36 (0.27–0.44)0.27 (0.18–0.36) Yield of identifying HGIN/EC (%)8484718686NOTE. In addition, this Table shows the yield of indentifying areas containing high-grade intraepithelial neoplasia (HGIN) or early cancer (EC) with WLE or combinations of WLE with NBI, ICC, or AAC or all 4 techniques together. Open table in a new tab NOTE. In addition, this Table shows the yield of indentifying areas containing high-grade intraepithelial neoplasia (HGIN) or early cancer (EC) with WLE or combinations of WLE with NBI, ICC, or AAC or all 4 techniques together. There was no difference in interobserver agreement between expert and nonexpert endoscopists, except for 3 evaluations. In the evaluation of the regularity of the mucosal pattern using only WLE images, the interobserver agreement of nonexpert endoscopists (κ = 0.64; 95% CI: 0.58–0.70) was better than that of expert endoscopists (κ = 0.43; 95% CI: 0.33–0.53). This was also found when the evaluation of this item was compared for the assessment with the combination of WLE plus NBI: interobserver agreement for nonexperts was κ = 0.51 (95% CI: 0.45–0.58) vs κ = 0.18 (95% CI: 0.07–0.28) for expert endoscopists. Finally, in the evaluation of all techniques together, nonexpert endoscopists had a lower interobserver agreement (κ = 0.27; 95% CI: 0.18–0.36) c
Referência(s)