- Email: [email protected]
422 Computer-Assisted Image Interpretation of Volumetric Laser Endomicroscopy in Barrett's Esophagus
Quantitative Analysis of Volumetric Laser Endomicroscopy of Histologically Correlated Images Potentially Identifies Early Neoplasia in Barrett's Esophagus Anne-Fré Swager, Daniel M. de Bruin, Dirk J. Faber, Bas L. Weusten, Sybren L. Meijer, Jacques J. Bergman, Ton G. van Leeuwen, Wouter L. Curvers Background: Early neoplastic lesions in Barrett's esophagus (BE) are difficult to detect with white-light endoscopy. Volumetric laser endomicroscopy (VLE) is an optical coherence tomography (OCT)-based imaging technique that provides large circumferential sub-surface maps of the superficial esophageal wall layers at low-power microscopy resolution. VLE data can be quantified by measuring the attenuation coefficient (µVLE), the decay of detected backscattered light versus depth. µVLE has the potential of providing quantitative optical diagnosis of interrogated mucosa because it relates to the organization of tissue. Aim: To investigate the feasibility of µVLE for identification of early neoplasia in BE. Methods: Endoscopic resection (ER) specimens from BE patients with and without neoplasia were scanned ex-vivo with VLE. Histopathology slides from the specimens were correlated oneto-one with VLE scans based on in-vivo and ex-vivo placed markers. Quantification of VLE signal attenuation (µVLE) was performed on areas of interest (AoIs) from VLE scans that were matched with histology in order to differentiate dysplastic BE from non-dysplastic (ND)BE mucosa. Results: In this study, 42 endoscopic resection (ER) specimens from 23 patients yielded 65 histology-VLE matches containing 70 AoIs consisting of 37 NDBE and 33 dysplastic BE AoIs (HGD n=15, EAC n=18). Median µVLE values (mm-1) of the different mucosa types were compared: NDBE 1.24 (0.40-1.73 IQR) and dysplastic BE 1.99 (0.892.81 IQR). A statistically significant difference was observed between these groups (p = 0.002). Conclusion: Quantitative VLE by means of µVLE may differentiate early neoplastic BE from NDBE. Further research is necessary to validate µVLE for diagnosis of early neoplasia in BE. Figure 1: Placement of a high resolution fiber optic catheter beyond the Ligament of Treitz of a healthy volunteer.
421 Development and Validation of a NBI Classification System for the Prediction of Dysplasia in Barrett's Esophagus (BE): Consensus Results From an International Working Group Benjamin R. Alsop, Jacques J. Bergman, Kenichi Goda, Irving Waxman, Mototsugu Kato, Helmut Messmann, Neil Gupta, Prashanth Vennalaganti, John R. Goldblum, Prateek Sharma Background: Although several criteria have been proposed for the evaluation of dysplasia in Barrett's Esophagus (BE) using NBI, none have been widely accepted. Aims: To develop and assess the validity of a consensus NBI classification of Barrett's esophagus. Methods: An international working group comprised of experts from USA, Europe and Japan was convened to develop standardized criteria for NBI in patients with Barrett's esophagus, the Barrett's International NBI Group (BING). The study was conducted in 3 phases. In phase 1, consensus meetings were held during which the BE experts evaluated NBI images showing characteristics of intestinal metaplasia/non dysplastic BE and high-grade dysplasia/cancer to arrive at a consensus classification system. In phase 2, using high definition NBI endoscopes (Olympus 190, near focus), high quality images of non-dysplastic and dysplastic BE areas were prospectively collected along with corresponding biopsies from BE patients. All biopsies were read in blinded fashion by a central expert GI pathologist. During the initial standardization, a web-based pretest of 50 NBI BE images was conducted in which all experts graded images using the BING classification. Finally, in phase 3, a total of 120 images, not previously viewed, were utilized to validate the criteria using a web-based survey. During the review of images, all experts applied the classification system to predict either "NDBE" or "HGD/ EAC", indicated the confidence level of the prediction and graded image quality. The overall accuracy, sensitivity, specificity, NPV and PPV of the assessors was calculated and k was determined for inter-observer agreement of predicted histology. Results: The BING classification system for BE utilized both the mucosal and vascular patterns seen within the BE mucosa and classified them as "regular" or "irregular" based on NBI features (straight/oval/ round/tubular mucosal patterns with fine/thin caliber blood vessels arranged in an orderly fashion were classified as regular BE patterns). Using the BING classification system, the overall accuracy of the experts in prediction of histology was 85%, with sensitivity 80% and specificity 88% (Table). There was a significant improvement in all parameters when the prediction was made with high confidence: accuracy 92%, sensitivity 91%, NPV 95%. The overall strength of inter-observer agreement in histology prediction was substantial, with a k of 0.68. Conclusions: The BING classification system has been developed by an international working group with expertise in BE. The classification system is simple, and evaluation of the mucosal and vascular patterns using this system can predict the presence or absence of dysplasia with high accuracy and NPV, especially in high confidence predictions. The inter observer agreement using this classification system is substantial.
Figure 2: Spatiotemporal plot of data recorded from the esophago-jejunal segment of a healthy volunteer in the build up to the onset of phase III MMC activity
419 Gastrointestinal Epithelia Could Be Luminal Sensors for Chemical, Hypo Osmolality, Acid, Temperature and Stretch via ATP Release Hiroshi Mihara, Jibran Sualeh Muhammad, Nobuhiro Suzuki, Yoshiaki Tabuchi, Toshiro Sugiyama Background/Purpose: Visceral hypersensitivity for acid, extension and temperature stimuli in stomach and duodenum are involved in functional dyspepsia (FD). While the involvement of TRPV1 and TRPA1 have been reported in a number of animal experiments, data remains conflicting. On the other hand, ATP-gated ion channels, especially P2X3 receptor have recently been worthy of attention for its role in visceral hypersensitivity. We have reported that TRPV4 is expressed in esophageal, gastric and intestinal epithelia (J Physiol 2011, DDW2013, Am J Physiol 2013) and that ATP exocytosis is induced by TRPV4 activation in esophageal keratinocytes. We hypothesized that various physiological stimuli induced ATP release from gastric and duodenal (intestinal) epithelia through each different receptor and that excessive ATP release was involved in the pathophysiology of visceral hypersensitivity. METHODS: TRPV4 expression was examined in normal rat gastric epithelial cell line (RGE), human gastric cancer cell line (AGS) and rat intestinal cell line (IEC-6) by RT-PCR, immunostaining or Western blotting. ATP releases responding to various physiological stimuli (chemicals, low osmotic pressure, acid, extension, temperature) from each cell line were examined using luciferin-luciferase reaction. Acidic (pH 4.0), warm (40 °C) and standard (25 °C, pH7.4) solutions were used as acidic, warm stimuli or a control, respectively. Stretch stimuli was applied to cells cultured on a silicon chamber using a stretching apparatus (STREX Inc, Osaka, Japan). RESULTS: TRPV4 was expressed in RGE, IEC-6 but not in AGS. Specific TRPV4 agonist (GSK1016790A) or endogenous TRPV4 activator (5,6-EET) significantly enhanced ATP release from RGE cells but not AGS. Low osmotic, acidic, stretch or temperature stimuli significantly enhanced ATP release in RGE or IEC-6 cells. Stretchinduced ATP release from RGE cells was inhibited by pretreatment with a specific TRPV4 inhibitor, HC067047. Cell viability after each stimulus was confirmed by Trypan blue staining. Conclusion: Gastric and duodenal (intestinal) epithelial cells release ATP responding various physiological stimuli partially via TRPV4 activation . Excessive ATP release from epithelial cells might be involved in the pathophysiology of visceral hypersensitivity.
422 Computer-Assisted Image Interpretation of Volumetric Laser Endomicroscopy in Barrett's Esophagus Eladio Rodriguez-Diaz, Satish K. Singh OBJECTIVE: Volumetric laser endomicroscopy (VLE) uses the principles of optical coherence tomography (OCT) to generate high resolution, grayscale, cross sectional images of tissue microstructures. When used in the esophagus, this approach reveals in real-time morphologies associated with Barrett's esophagus (BE) and dysplasia. However, current approaches
S-91
X : 55624$1AGA 03-28-15 04:55:56 PDFd : 55624B : o
Page 91
AGA Abstracts
AGA Abstracts
420
AGA Abstracts
using VLE for surveillance of dysplasia associated with BE require the interpretation of large volumes of VLE image data by an expert to identify pathologies of interest. The main objective of the current study is to explore the feasibility of developing a diagnostic computer algorithm to assist in the interpretation of VLE image data. METHODS: Retrospective data analysis study. De-identified VLE images from the esophagus obtained with the NvisionVLE radial OCT system (NinePoint Medical, Cambridge, MA), along with their corresponding validated pathology classifications were used in the study. We used the discrete wavelet transform to analyze and characterize the local spatial distribution of gray levels in the VLE images. Three classification problems relevant to the current clinical use of VLE in BE were considered: distinguishing gastric cardia (GC) from BE (analogous to determining the location of the esophagogastric junction), normal squamous (NS) from BE (finding the top of the BE segment), and non-dysplastic BE (NDBE) from BE with dysplasia (BED). Features extracted from the wavelet components that were found to be statistically significant based on a 2sample t-test were used as inputs to a Naïve Bayes classifier. Leave-one-out cross-validation was used to obtain performance estimates for each of the classification problems. RESULTS: A total of 60 images (18 from NS, 15 from NDBE, 14 from BED, and 13 from GC) from 38 patients were analyzed. Distinguishing GC from BE (NDBE as well as BED) resulted in per-class accuracies of 77% and 93% respectively. Distinguishing NS from BE resulted in per-class accuracies of 83% for both classes. A sensitivity of 86% and specificity of 93% were obtained when distinguishing NDBE from BED. CONCLUSION: Real-time identification of potential areas of dysplasia by VLE imaging can help to better target biopsies when compared to the standard protocol of random biopsies. Computer-assisted interpretation of VLE images may provide a novel approach to making objective, real-time assessments of pathologies relevant to dysplasia surveillance in BE. Our results offer promise for the feasibility of such an approach, where VLE images could be instantaneously analyzed and interpreted on-the-fly to guide endoscopic diagnosis and treatment.
cost of the two strategies from a public health payer perspective. CLE was priced as a therapeutic endoscopy procedure that is valued 20% more than a standard diagnostic endoscopy. RESULTS: In spite of a higher procedure cost, the total costs for a cohort of 100 patients associated to the CLE-based strategy were 89,313.29 ¤ compared to 90,658.90 ¤ for the standard strategy. The cost per patient adequately diagnosed and treated was also inferior for the CLE-based strategy (960.36 ¤ vs. 1,054.17 ¤). Different sensitivity analyses (including a Monte-Carlo modeling) were performed which confirmed the robustness of previous results. CONCLUSIONS: In the restricted context of this evaluation, the CLEbased strategy is not only more effective but also less costly than the standard strategy i.e. corresponding to the definition of a dominant strategy. REFERENCES: 1. Canto MI, Anandasabapathy S, Brugge W, Falk GW, Dunbar KB, Zhang Z, Woods K,Almario JA, Schell U, Goldblum J, Maitra A, Montgomery E, Kiesslich R; Confocal Endomicroscopy for Barrett's Esophagus or Confocal Endomicroscopy for Barrett's Esophagus (CEBE) Trial Group. In vivo endomicroscopy improves detection of Barrett's esophagus-related neoplasia: a multicenter international randomized controlled trial (with video). Gastrointest Endosc.2014 Feb;79(2):211-21. doi: 10.1016/ j.gie.2013.09.020. Epub 2013 Nov 9.
425 Neosquamous Esophageal Epithelium Can Be Differentiated From Native Esophageal Squamous Epithelium Using Volumetric Laser Endomicroscopy (VLE) Tarun Rai, Donald R. Campbell, Prashanth Vennalaganti, Benjamin R. Alsop, Kelsey L. Able, Mohammad A. Titi, Neil Gupta, Abhishek Choudhary, Ajay Bansal, Prateek Sharma Background: After endoscopic eradication therapy of neoplastic Barrett's esophagus (BE), the columnar epithelium is replaced by stratified squamous epithelium also known as neo-squamous epithelium (NeSE). NeSE is endoscopically indistinguishable from native squamous epithelium (NaSE). VLE, a novel imaging technique based on optical frequencydomain imaging (OFDI) has been shown to differentiate BE from NaSE but its role in differentiating NeSE from NaSE is not known. Aim: To determine VLE criteria that could potentially distinguish between neo and native squamous esophageal epithelium of the esophagus Methods: VLE images (NVision VLE Imaging System, NinePoint Medical) of patients with BE that had been prospectively obtained as part of a multi-center VLE study were retrieved. Images of both native squamous epithelium (above the columnar lined esophagus of non-ablated BE patients) and neo-squamous epithelium (distal esophagus of those who had undergone endoscopic ablation of high-grade dysplasia /EAC) were identified. All images were reviewed by 3 investigators including an experienced VLE endoscopist to determine VLE criteria that could potentially distinguish between neo and native squamous esophageal epithelium of the esophagus. Results: VLE images obtained from 43 patients were prospectively evaluated; 26 with NaSE (without prior history of endoscopic intervention) and 17 with NeSE (post-endoscopic mucosal resection and/or radiofrequency ablation). Images from 6 patients were excluded from the final analysis; 5 patients had were poor quality images whereas another patient had history of radiation therapy to the esophagus. In VLE images from the NaSE, discrete surface layers were identified with minimal or no glandular structures whereas in images from the NeSE. the surface layers were discontinuous. Further detailed analysis of the VLE images revealed the following criteria for scoring: Surface layers were assigned a value of 2 for presence of discrete layers, 1 for discontinuous and 0 for absence of layered architecture. Glands were assigned values of -2 if present in abundance, -1 if occasionally present and 0 if they were absent. Score values (sum for surface layers plus glands) were as follows for the different epithelia: NaSE =2; NeSE = 0 and BE = -2. Conclusion: We have reported the development of simple VLE criteria that can potentially differentiate between neo-squamous and native squamous epithelium. If these criteria are validated, they may prove to be clinically useful in the surveillance of patients after endoscopic therapy for neoplastic BE.
423 Can In-Vivo Volumetric Laser Endomicroscopy Detect Barrett's Esophagus Dysplasia Missed on Surveillance Biopsies? Cadman L. Leggett, Daniel K. Chan, Emmanuel C. Gorospe, Akhil Nehra, Marlys Anderson, Lori S. Lutzke, Kenneth K. Wang Background: Volumetric laser endomicroscopy (VLE) is a second-generation optical coherence tomography device used in detecting dysplasia associated with Barrett's esophagus (BE). The diagnostic performance of in-vivo VLE has been studied by comparing VLE ratings to overall histology. Precise in-vivo histology to image correlation is technically challenging, limiting the ability to determine whether VLE can detect dysplasia missed on surveillance biopsies. Aim: To determine whether VLE can detect BE dysplasia missed at index endoscopy (EGD) Methods: 16 patients with a history of dysplastic BE underwent in-vivo VLE. All patients had a history of endoscopic therapy and were required to have ‡2 post-VLE surveillance EGDs. VLE scans were independently rated by two reviewers with experience in VLE using a previously described diagnostic algorithm. Reviewers were blinded to histology and rating consensus was reached by review of disparate ratings. Scans were rated as neoplastic if they contained complete effacement of the mucosal layer with surface>subsurface intensity or partial effacement of the mucosal layer with >5 atypical glands. Scans were rated as non-neoplastic if they contained squamous epithelium (SE), complete effacement of the mucosal layer with surface£subsurface intensity or partial effacement of the mucosal layer with £5 atypical glands. Highest grade histology was recorded at time of VLE and for all post-VLE surveillance EGDs. Histology was dichotomized into non-neoplastic (SE, nondysplastic BE, low-grade dysplasia) and neoplastic (high-grade dysplasia, intramucosal adenocarcinoma). VLE ratings were compared to histology at time of VLE and to highest grade histology across all post-VLE surveillance EGDs. Results: Study participants had a mean (SD) age of 65 (7) years and 14 (87%) were male. 16 (100%) patients had a history of endoscopic resection and 14 (87%) of radiofrequency ablation. The mean (SD) BE length was 3(3) cm. Patients underwent a mean (SD) of 4(2) post-VLE surveillance EGDs. A total of 7 (44%) scans were rated as neoplastic. At VLE, 2(13%) EGDs showed neoplastic histology. On post-VLE surveillance, 4(25%) EGD's showed neoplastic histology. 2 VLE scans rated as neoplastic with non-neoplastic histology at VLE showed evidence of neoplasia on postVLE surveillance EGDs, corresponding to a two-fold increase in dysplasia detection among VLE scans rated as neoplastic: at VLE (N=2, 28%) vs. post-VLE (N=4, 57%). 9 (56%) scans rated as non-neoplastic showed non-neoplastic histology at VLE EGD and on post-VLE EGDs. Conclusion: In-vivo VLE can potentially detect dysplasia missed on surveillance biopsies. Our study is unable to determine whether this was secondary to dysplasia missed with biopsy or progression of disease although this is of no clinically significant difference.
443 Association of GPBAR1 Gene Variant With Satiation and Circulating Fibroblast Growth Factor-19 Supports a Role of Bile Acids in Obesity Andres Acosta, Nelson Valentin Feliciano, Paula Carlson, Deborah J. Eckert, Jessica O'Neill, Michael Camilleri Background: Recent studies showed that bariatric surgery results in doubling of the concentration of serum bile acids (BA), and increases serum adiponectin, circulating fibroblast growth factor-19 (FGF-19) and GLP-1 (Nature 2014;509(7499):183-8). These changes are associated with improvement of obesity and diabetes. The BA effects may be mediated by the BA receptor GPBAR1 (also known as TGR-5) and the stimulation of farnesoid X receptor (FXR) in ileal L cells. We have previously shown that in patients with IBS, GPBAR1 rs11554825 CC genotype is associated with faster colonic transit and increased total fecal BA excretion (Am J Physiol 2014;307:G508-16), consistent with a gain of function. In order to further assess the role of GPBAR1, we tested the hypothesis that GPBAR1 gene variant is associated with altered FXR function, manifested by reduced postprandial FGF-19, and that GPBAR1 variation is associated with lower BMI, satiation, satiety, and changes in gastric function and gut hormones. Methods: 274 overweight or obese participants [age (SE) 37.4±0.7y, 70% females, predominantly Caucasian, BMI 32.9±0.3kg/m2] underwent, on separate days, validated measurements of GI endophenotypes: gastric emptying (GE) of solids and liquids by scintigraphy, gastric volume (GV) by SPECT during fasting and after 200mL Ensure® (D GV), satiety by kcal intake (T-kcal) during a buffet meal, satiation [volume to fullness (VTF) and maximal tolerated volume (MTV)] by Ensure® ingested at 30ml/min, and fasting and postprandial (15, 45 and 90 min) FGF-19, GLP-1, PYY, CCK, ghrelin (by radioimmunoassay). GPBAR1 gene variant rs11554825 was genotyped with TaqMan® SNP Assay. We assessed the association of GPBAR1 with each variable using ANOVA by general (TT vs. CT vs. CC) genetic model. Results: GPBAR1 rs11554825 CC gene variant was associated with numerically lower BMI (mean D 1.6kg/m2, p=0.07) and body weight (mean D 3.3kg, p=0.11) when compared to CT or TT. GPBAR1 rs11554825 CC genotype was associated with: 1) decreased calorie intake in satiation drink test (MTV mean D 158ml, p=0.03); 2) increased postprandial ghrelin (mean D at 90min, 6.8pg/ml, p=0.005); 3) increased postprandial FGF-19 AUC (mean D 71.5pg/ml, p=0.03); 4) decreased
424 Confocal Laser Endomicroscopy (CLE): A Dominant Strategy for the Management of Barrett's Esophagus Claude Le Pen, Emmanuel Coron, Jean Paul Galmiche INTRODUCTION: The diagnostic value of CLE in Barrett's Esophagus (BE) and its impact on the management of patients undergoing surveillance or treatment were recently established by Canto et al (1) in a prospective randomized trial. How these results can be extrapolated to a European healthcare economic model remains presently unknown. OBJECTIVES: To evaluate the cost-effectiveness of a CLE-based strategy compared to the standard Seattle protocol for the management of patients with BE and suspicion of neoplasia, in the setting of a French public healthcare system. METHODS The comparison of the CLE-based strategy to the standard Seattle Protocol used the data published by Canto et al (1) who evaluated 2 strategies: (i) high definition white light endoscopy (HD-WLE) + CLE and targeted physical biopsies (CLE-based strategy) and (ii) HD-WLE and random physical biopsies (Standard strategy). In that study, the CLE-based strategy had a higher sensitivity for the diagnosis of HGD/EC (95 versus 40%) without significant change in specificity (92% versus 98 % respectively). The average number of biopsies performed was reduced from 5.91 to 1.26 per patient by the use of CLE. These data were entered into a health economics model (piggyback study) to compare the costs of both strategies when performed in public academic hospitals. French costs were associated to each procedure in order to estimate the medical
AGA Abstracts
X : 55624$1AGA 03-28-15 04:55:56 PDFd : 55624B : e
S-92
Page 92
421 Development and Validation of a NBI Classification System for the Prediction of Dysplasia in Barrett's Esophagus (BE): Consensus Results From an International Working Group Benjamin R. Alsop, Jacques J. Bergman, Kenichi Goda, Irving Waxman, Mototsugu Kato, Helmut Messmann, Neil Gupta, Prashanth Vennalaganti, John R. Goldblum, Prateek Sharma Background: Although several criteria have been proposed for the evaluation of dysplasia in Barrett's Esophagus (BE) using NBI, none have been widely accepted. Aims: To develop and assess the validity of a consensus NBI classification of Barrett's esophagus. Methods: An international working group comprised of experts from USA, Europe and Japan was convened to develop standardized criteria for NBI in patients with Barrett's esophagus, the Barrett's International NBI Group (BING). The study was conducted in 3 phases. In phase 1, consensus meetings were held during which the BE experts evaluated NBI images showing characteristics of intestinal metaplasia/non dysplastic BE and high-grade dysplasia/cancer to arrive at a consensus classification system. In phase 2, using high definition NBI endoscopes (Olympus 190, near focus), high quality images of non-dysplastic and dysplastic BE areas were prospectively collected along with corresponding biopsies from BE patients. All biopsies were read in blinded fashion by a central expert GI pathologist. During the initial standardization, a web-based pretest of 50 NBI BE images was conducted in which all experts graded images using the BING classification. Finally, in phase 3, a total of 120 images, not previously viewed, were utilized to validate the criteria using a web-based survey. During the review of images, all experts applied the classification system to predict either "NDBE" or "HGD/ EAC", indicated the confidence level of the prediction and graded image quality. The overall accuracy, sensitivity, specificity, NPV and PPV of the assessors was calculated and k was determined for inter-observer agreement of predicted histology. Results: The BING classification system for BE utilized both the mucosal and vascular patterns seen within the BE mucosa and classified them as "regular" or "irregular" based on NBI features (straight/oval/ round/tubular mucosal patterns with fine/thin caliber blood vessels arranged in an orderly fashion were classified as regular BE patterns). Using the BING classification system, the overall accuracy of the experts in prediction of histology was 85%, with sensitivity 80% and specificity 88% (Table). There was a significant improvement in all parameters when the prediction was made with high confidence: accuracy 92%, sensitivity 91%, NPV 95%. The overall strength of inter-observer agreement in histology prediction was substantial, with a k of 0.68. Conclusions: The BING classification system has been developed by an international working group with expertise in BE. The classification system is simple, and evaluation of the mucosal and vascular patterns using this system can predict the presence or absence of dysplasia with high accuracy and NPV, especially in high confidence predictions. The inter observer agreement using this classification system is substantial.
Figure 2: Spatiotemporal plot of data recorded from the esophago-jejunal segment of a healthy volunteer in the build up to the onset of phase III MMC activity
419 Gastrointestinal Epithelia Could Be Luminal Sensors for Chemical, Hypo Osmolality, Acid, Temperature and Stretch via ATP Release Hiroshi Mihara, Jibran Sualeh Muhammad, Nobuhiro Suzuki, Yoshiaki Tabuchi, Toshiro Sugiyama Background/Purpose: Visceral hypersensitivity for acid, extension and temperature stimuli in stomach and duodenum are involved in functional dyspepsia (FD). While the involvement of TRPV1 and TRPA1 have been reported in a number of animal experiments, data remains conflicting. On the other hand, ATP-gated ion channels, especially P2X3 receptor have recently been worthy of attention for its role in visceral hypersensitivity. We have reported that TRPV4 is expressed in esophageal, gastric and intestinal epithelia (J Physiol 2011, DDW2013, Am J Physiol 2013) and that ATP exocytosis is induced by TRPV4 activation in esophageal keratinocytes. We hypothesized that various physiological stimuli induced ATP release from gastric and duodenal (intestinal) epithelia through each different receptor and that excessive ATP release was involved in the pathophysiology of visceral hypersensitivity. METHODS: TRPV4 expression was examined in normal rat gastric epithelial cell line (RGE), human gastric cancer cell line (AGS) and rat intestinal cell line (IEC-6) by RT-PCR, immunostaining or Western blotting. ATP releases responding to various physiological stimuli (chemicals, low osmotic pressure, acid, extension, temperature) from each cell line were examined using luciferin-luciferase reaction. Acidic (pH 4.0), warm (40 °C) and standard (25 °C, pH7.4) solutions were used as acidic, warm stimuli or a control, respectively. Stretch stimuli was applied to cells cultured on a silicon chamber using a stretching apparatus (STREX Inc, Osaka, Japan). RESULTS: TRPV4 was expressed in RGE, IEC-6 but not in AGS. Specific TRPV4 agonist (GSK1016790A) or endogenous TRPV4 activator (5,6-EET) significantly enhanced ATP release from RGE cells but not AGS. Low osmotic, acidic, stretch or temperature stimuli significantly enhanced ATP release in RGE or IEC-6 cells. Stretchinduced ATP release from RGE cells was inhibited by pretreatment with a specific TRPV4 inhibitor, HC067047. Cell viability after each stimulus was confirmed by Trypan blue staining. Conclusion: Gastric and duodenal (intestinal) epithelial cells release ATP responding various physiological stimuli partially via TRPV4 activation . Excessive ATP release from epithelial cells might be involved in the pathophysiology of visceral hypersensitivity.
422 Computer-Assisted Image Interpretation of Volumetric Laser Endomicroscopy in Barrett's Esophagus Eladio Rodriguez-Diaz, Satish K. Singh OBJECTIVE: Volumetric laser endomicroscopy (VLE) uses the principles of optical coherence tomography (OCT) to generate high resolution, grayscale, cross sectional images of tissue microstructures. When used in the esophagus, this approach reveals in real-time morphologies associated with Barrett's esophagus (BE) and dysplasia. However, current approaches
S-91
X : 55624$1AGA 03-28-15 04:55:56 PDFd : 55624B : o
Page 91
AGA Abstracts
AGA Abstracts
420
AGA Abstracts
using VLE for surveillance of dysplasia associated with BE require the interpretation of large volumes of VLE image data by an expert to identify pathologies of interest. The main objective of the current study is to explore the feasibility of developing a diagnostic computer algorithm to assist in the interpretation of VLE image data. METHODS: Retrospective data analysis study. De-identified VLE images from the esophagus obtained with the NvisionVLE radial OCT system (NinePoint Medical, Cambridge, MA), along with their corresponding validated pathology classifications were used in the study. We used the discrete wavelet transform to analyze and characterize the local spatial distribution of gray levels in the VLE images. Three classification problems relevant to the current clinical use of VLE in BE were considered: distinguishing gastric cardia (GC) from BE (analogous to determining the location of the esophagogastric junction), normal squamous (NS) from BE (finding the top of the BE segment), and non-dysplastic BE (NDBE) from BE with dysplasia (BED). Features extracted from the wavelet components that were found to be statistically significant based on a 2sample t-test were used as inputs to a Naïve Bayes classifier. Leave-one-out cross-validation was used to obtain performance estimates for each of the classification problems. RESULTS: A total of 60 images (18 from NS, 15 from NDBE, 14 from BED, and 13 from GC) from 38 patients were analyzed. Distinguishing GC from BE (NDBE as well as BED) resulted in per-class accuracies of 77% and 93% respectively. Distinguishing NS from BE resulted in per-class accuracies of 83% for both classes. A sensitivity of 86% and specificity of 93% were obtained when distinguishing NDBE from BED. CONCLUSION: Real-time identification of potential areas of dysplasia by VLE imaging can help to better target biopsies when compared to the standard protocol of random biopsies. Computer-assisted interpretation of VLE images may provide a novel approach to making objective, real-time assessments of pathologies relevant to dysplasia surveillance in BE. Our results offer promise for the feasibility of such an approach, where VLE images could be instantaneously analyzed and interpreted on-the-fly to guide endoscopic diagnosis and treatment.
cost of the two strategies from a public health payer perspective. CLE was priced as a therapeutic endoscopy procedure that is valued 20% more than a standard diagnostic endoscopy. RESULTS: In spite of a higher procedure cost, the total costs for a cohort of 100 patients associated to the CLE-based strategy were 89,313.29 ¤ compared to 90,658.90 ¤ for the standard strategy. The cost per patient adequately diagnosed and treated was also inferior for the CLE-based strategy (960.36 ¤ vs. 1,054.17 ¤). Different sensitivity analyses (including a Monte-Carlo modeling) were performed which confirmed the robustness of previous results. CONCLUSIONS: In the restricted context of this evaluation, the CLEbased strategy is not only more effective but also less costly than the standard strategy i.e. corresponding to the definition of a dominant strategy. REFERENCES: 1. Canto MI, Anandasabapathy S, Brugge W, Falk GW, Dunbar KB, Zhang Z, Woods K,Almario JA, Schell U, Goldblum J, Maitra A, Montgomery E, Kiesslich R; Confocal Endomicroscopy for Barrett's Esophagus or Confocal Endomicroscopy for Barrett's Esophagus (CEBE) Trial Group. In vivo endomicroscopy improves detection of Barrett's esophagus-related neoplasia: a multicenter international randomized controlled trial (with video). Gastrointest Endosc.2014 Feb;79(2):211-21. doi: 10.1016/ j.gie.2013.09.020. Epub 2013 Nov 9.
425 Neosquamous Esophageal Epithelium Can Be Differentiated From Native Esophageal Squamous Epithelium Using Volumetric Laser Endomicroscopy (VLE) Tarun Rai, Donald R. Campbell, Prashanth Vennalaganti, Benjamin R. Alsop, Kelsey L. Able, Mohammad A. Titi, Neil Gupta, Abhishek Choudhary, Ajay Bansal, Prateek Sharma Background: After endoscopic eradication therapy of neoplastic Barrett's esophagus (BE), the columnar epithelium is replaced by stratified squamous epithelium also known as neo-squamous epithelium (NeSE). NeSE is endoscopically indistinguishable from native squamous epithelium (NaSE). VLE, a novel imaging technique based on optical frequencydomain imaging (OFDI) has been shown to differentiate BE from NaSE but its role in differentiating NeSE from NaSE is not known. Aim: To determine VLE criteria that could potentially distinguish between neo and native squamous esophageal epithelium of the esophagus Methods: VLE images (NVision VLE Imaging System, NinePoint Medical) of patients with BE that had been prospectively obtained as part of a multi-center VLE study were retrieved. Images of both native squamous epithelium (above the columnar lined esophagus of non-ablated BE patients) and neo-squamous epithelium (distal esophagus of those who had undergone endoscopic ablation of high-grade dysplasia /EAC) were identified. All images were reviewed by 3 investigators including an experienced VLE endoscopist to determine VLE criteria that could potentially distinguish between neo and native squamous esophageal epithelium of the esophagus. Results: VLE images obtained from 43 patients were prospectively evaluated; 26 with NaSE (without prior history of endoscopic intervention) and 17 with NeSE (post-endoscopic mucosal resection and/or radiofrequency ablation). Images from 6 patients were excluded from the final analysis; 5 patients had were poor quality images whereas another patient had history of radiation therapy to the esophagus. In VLE images from the NaSE, discrete surface layers were identified with minimal or no glandular structures whereas in images from the NeSE. the surface layers were discontinuous. Further detailed analysis of the VLE images revealed the following criteria for scoring: Surface layers were assigned a value of 2 for presence of discrete layers, 1 for discontinuous and 0 for absence of layered architecture. Glands were assigned values of -2 if present in abundance, -1 if occasionally present and 0 if they were absent. Score values (sum for surface layers plus glands) were as follows for the different epithelia: NaSE =2; NeSE = 0 and BE = -2. Conclusion: We have reported the development of simple VLE criteria that can potentially differentiate between neo-squamous and native squamous epithelium. If these criteria are validated, they may prove to be clinically useful in the surveillance of patients after endoscopic therapy for neoplastic BE.
423 Can In-Vivo Volumetric Laser Endomicroscopy Detect Barrett's Esophagus Dysplasia Missed on Surveillance Biopsies? Cadman L. Leggett, Daniel K. Chan, Emmanuel C. Gorospe, Akhil Nehra, Marlys Anderson, Lori S. Lutzke, Kenneth K. Wang Background: Volumetric laser endomicroscopy (VLE) is a second-generation optical coherence tomography device used in detecting dysplasia associated with Barrett's esophagus (BE). The diagnostic performance of in-vivo VLE has been studied by comparing VLE ratings to overall histology. Precise in-vivo histology to image correlation is technically challenging, limiting the ability to determine whether VLE can detect dysplasia missed on surveillance biopsies. Aim: To determine whether VLE can detect BE dysplasia missed at index endoscopy (EGD) Methods: 16 patients with a history of dysplastic BE underwent in-vivo VLE. All patients had a history of endoscopic therapy and were required to have ‡2 post-VLE surveillance EGDs. VLE scans were independently rated by two reviewers with experience in VLE using a previously described diagnostic algorithm. Reviewers were blinded to histology and rating consensus was reached by review of disparate ratings. Scans were rated as neoplastic if they contained complete effacement of the mucosal layer with surface>subsurface intensity or partial effacement of the mucosal layer with >5 atypical glands. Scans were rated as non-neoplastic if they contained squamous epithelium (SE), complete effacement of the mucosal layer with surface£subsurface intensity or partial effacement of the mucosal layer with £5 atypical glands. Highest grade histology was recorded at time of VLE and for all post-VLE surveillance EGDs. Histology was dichotomized into non-neoplastic (SE, nondysplastic BE, low-grade dysplasia) and neoplastic (high-grade dysplasia, intramucosal adenocarcinoma). VLE ratings were compared to histology at time of VLE and to highest grade histology across all post-VLE surveillance EGDs. Results: Study participants had a mean (SD) age of 65 (7) years and 14 (87%) were male. 16 (100%) patients had a history of endoscopic resection and 14 (87%) of radiofrequency ablation. The mean (SD) BE length was 3(3) cm. Patients underwent a mean (SD) of 4(2) post-VLE surveillance EGDs. A total of 7 (44%) scans were rated as neoplastic. At VLE, 2(13%) EGDs showed neoplastic histology. On post-VLE surveillance, 4(25%) EGD's showed neoplastic histology. 2 VLE scans rated as neoplastic with non-neoplastic histology at VLE showed evidence of neoplasia on postVLE surveillance EGDs, corresponding to a two-fold increase in dysplasia detection among VLE scans rated as neoplastic: at VLE (N=2, 28%) vs. post-VLE (N=4, 57%). 9 (56%) scans rated as non-neoplastic showed non-neoplastic histology at VLE EGD and on post-VLE EGDs. Conclusion: In-vivo VLE can potentially detect dysplasia missed on surveillance biopsies. Our study is unable to determine whether this was secondary to dysplasia missed with biopsy or progression of disease although this is of no clinically significant difference.
443 Association of GPBAR1 Gene Variant With Satiation and Circulating Fibroblast Growth Factor-19 Supports a Role of Bile Acids in Obesity Andres Acosta, Nelson Valentin Feliciano, Paula Carlson, Deborah J. Eckert, Jessica O'Neill, Michael Camilleri Background: Recent studies showed that bariatric surgery results in doubling of the concentration of serum bile acids (BA), and increases serum adiponectin, circulating fibroblast growth factor-19 (FGF-19) and GLP-1 (Nature 2014;509(7499):183-8). These changes are associated with improvement of obesity and diabetes. The BA effects may be mediated by the BA receptor GPBAR1 (also known as TGR-5) and the stimulation of farnesoid X receptor (FXR) in ileal L cells. We have previously shown that in patients with IBS, GPBAR1 rs11554825 CC genotype is associated with faster colonic transit and increased total fecal BA excretion (Am J Physiol 2014;307:G508-16), consistent with a gain of function. In order to further assess the role of GPBAR1, we tested the hypothesis that GPBAR1 gene variant is associated with altered FXR function, manifested by reduced postprandial FGF-19, and that GPBAR1 variation is associated with lower BMI, satiation, satiety, and changes in gastric function and gut hormones. Methods: 274 overweight or obese participants [age (SE) 37.4±0.7y, 70% females, predominantly Caucasian, BMI 32.9±0.3kg/m2] underwent, on separate days, validated measurements of GI endophenotypes: gastric emptying (GE) of solids and liquids by scintigraphy, gastric volume (GV) by SPECT during fasting and after 200mL Ensure® (D GV), satiety by kcal intake (T-kcal) during a buffet meal, satiation [volume to fullness (VTF) and maximal tolerated volume (MTV)] by Ensure® ingested at 30ml/min, and fasting and postprandial (15, 45 and 90 min) FGF-19, GLP-1, PYY, CCK, ghrelin (by radioimmunoassay). GPBAR1 gene variant rs11554825 was genotyped with TaqMan® SNP Assay. We assessed the association of GPBAR1 with each variable using ANOVA by general (TT vs. CT vs. CC) genetic model. Results: GPBAR1 rs11554825 CC gene variant was associated with numerically lower BMI (mean D 1.6kg/m2, p=0.07) and body weight (mean D 3.3kg, p=0.11) when compared to CT or TT. GPBAR1 rs11554825 CC genotype was associated with: 1) decreased calorie intake in satiation drink test (MTV mean D 158ml, p=0.03); 2) increased postprandial ghrelin (mean D at 90min, 6.8pg/ml, p=0.005); 3) increased postprandial FGF-19 AUC (mean D 71.5pg/ml, p=0.03); 4) decreased
424 Confocal Laser Endomicroscopy (CLE): A Dominant Strategy for the Management of Barrett's Esophagus Claude Le Pen, Emmanuel Coron, Jean Paul Galmiche INTRODUCTION: The diagnostic value of CLE in Barrett's Esophagus (BE) and its impact on the management of patients undergoing surveillance or treatment were recently established by Canto et al (1) in a prospective randomized trial. How these results can be extrapolated to a European healthcare economic model remains presently unknown. OBJECTIVES: To evaluate the cost-effectiveness of a CLE-based strategy compared to the standard Seattle protocol for the management of patients with BE and suspicion of neoplasia, in the setting of a French public healthcare system. METHODS The comparison of the CLE-based strategy to the standard Seattle Protocol used the data published by Canto et al (1) who evaluated 2 strategies: (i) high definition white light endoscopy (HD-WLE) + CLE and targeted physical biopsies (CLE-based strategy) and (ii) HD-WLE and random physical biopsies (Standard strategy). In that study, the CLE-based strategy had a higher sensitivity for the diagnosis of HGD/EC (95 versus 40%) without significant change in specificity (92% versus 98 % respectively). The average number of biopsies performed was reduced from 5.91 to 1.26 per patient by the use of CLE. These data were entered into a health economics model (piggyback study) to compare the costs of both strategies when performed in public academic hospitals. French costs were associated to each procedure in order to estimate the medical
AGA Abstracts
X : 55624$1AGA 03-28-15 04:55:56 PDFd : 55624B : e
S-92
Page 92