Search

ObjectiveThe diagnosis of autoimmune pancreatitis (AIP) is challenging. Sonographic and cross-sectional imaging findings of AIP closely mimic pancreatic ductal adenocarcinoma (PDAC) and techniques for tissue sampling of AIP are suboptimal. These limitations often result in delayed or failed diagnosis, which negatively impact patient management and outcomes. This study aimed to create an endoscopic ultrasound (EUS)-based convolutional neural network (CNN) model trained to differentiate AIP from PDAC, chronic pancreatitis (CP) and normal pancreas (NP), with sufficient performance to analyse EUS video in real time.DesignA database of still image and video data obtained from EUS examinations of cases of AIP, PDAC, CP and NP was used to develop a CNN. Occlusion heatmap analysis was used to identify sonographic features the CNN valued when differentiating AIP from PDAC.ResultsFrom 583 patients (146 AIP, 292 PDAC, 72 CP and 73 NP), a total of 1 174 461 unique EUS images were extracted. For video data, the CNN processed 955 EUS frames per second and was: 99% sensitive, 98% specific for distinguishing AIP from NP; 94% sensitive, 71% specific for distinguishing AIP from CP; 90% sensitive, 93% specific for distinguishing AIP from PDAC; and 90% sensitive, 85% specific for distinguishing AIP from all studied conditions (ie, PDAC, CP and NP).ConclusionThe developed EUS-CNN model accurately differentiated AIP from PDAC and benign pancreatic conditions, thereby offering the capability of earlier and more accurate diagnosis. Use of this model offers the potential for more timely and appropriate patient care and improved outcome.

SUMMARY Given their malignant potential, resection of esophageal granular cell tumors (GCTs) is often undertaken, yet the optimal technique is unknown. We present a large series of dedicated endoscopic resection using band ligation (EMR-B) of esophageal GCTs. Patients diagnosed with esophageal GCTs between 2002 and 2019 were identified using a prospectively collected pathology database. Endoscopic reports were reviewed, and patients who underwent dedicated EMR-B of esophageal GCTs were included. Medical records were queried for demographics, findings, adverse events, and follow-up. We identified 21 patients who underwent dedicated EMR-B for previously identified esophageal GCT. Median age was 39 years; 16 (76%) were female. Eight (38%) had preceding signs or symptoms, potentially attributable to the GCT. Upon endoscopic evaluation, 12 (57%) were found in the distal esophagus. Endoscopic ultrasound was used in 15 cases (71%). Median lesion size was 7 mm, interquartile range 4 mm—8 mm. The largest lesion was 12 mm. A total of 20 (95%) had en bloc resection confirmed with pathologic examination. The only patient with tumor extending to the resection margin underwent surveillance endoscopy that showed no residual tumor. No patients experienced bleeding, perforation, or stricturing in our series. No patients have had known recurrence of their esophageal GCT. EMR-B of esophageal GCT achieves complete histopathologic resection with minimal adverse events. EMR-B is safe and effective and seems prudent compared with observation for what could be an aggressive and malignant tumor. EMR-B should be considered first-line therapy when resecting esophageal GCT up to 12 mm in diameter.

Background and Aims Pancreatic necrosis may be categorized as an acute necrotic collection (ANC) or walled-off necrosis (WON) based on complete encapsulation by a wall and collection age (≤4 weeks or >4 weeks). Endoscopic intervention of WON has become the standard of care, but little is known regarding the safety and efficacy of endoscopic intervention of pancreatic necrosis ≤4 weeks from disease onset. Methods Retrospective review of medical records and imaging studies of all patients undergoing early endoscopic intervention of pancreatic necrosis between 2008 and 2018 was carried out at 1 referral center. Patients who underwent previous interventional treatment were excluded. Control WON patients were matched to early intervention cases. The primary outcome was defined as resolution of the collection after endoscopic treatment, without surgery. Results Nineteen patients with early intervention were identified. The most common indication for intervention was infection. Median age of these collections at the time of initial endoscopic intervention was 23 days (range, 15-27 days), and all collections had a partial or complete wall discernable on contrast-enhanced CT. Eleven patients underwent concurrent endoscopic necrosectomy. The primary outcome was achieved in all patients in the early intervention group. Total duration of therapy was longer for early intervention compared with controls (103 vs 69 days, P = .042), with no mortality and similar adverse event rates compared with controls. Conclusions Endoscopic intervention of pancreatic necrosis in the third and fourth weeks of illness appears effective and safe when a partial collection wall is present on cross-sectional imaging studies, with outcomes paralleling those reported for intervention of WON.

To investigate advanced neoplasia (AN) after endoscopic mucosal resection (EMR) of colonic laterally spreading lesions (LSLs). A retrospective study of patients who underwent injection-assisted EMR of colonic LSLs ≥ 10 mm was performed. Primary outcome was overall rate of AN at initial surveillance colonoscopy. Secondary outcomes were the rates of residual AN (rAN) at the EMR site and metachronous AN (mAN), and analysis of risk factors for AN, including effect of surveillance guidance. Three hundred seventy-four patients underwent successful EMR for 388 LSLs. AN occurred in 66/374 (17.6%) patients on initial surveillance colonoscopy at median follow-up of 364.5 days. Two patients had both rAN and mAN, for a total of 68 instances of AN, including 30/374 (8.0%) cases of rAN and 38/374 (10.2%) cases of mAN. On multivariate analysis, use of piecemeal resection was associated with increased likelihood of residual AN (P = 0.003, OR 9.2, 95% CI 2.1–33.3). Twenty-nine out of thirty cases (96.7%) of rAN were successfully endoscopically managed at surveillance colonoscopy. AN occurred in 17.6% of all patients at initial surveillance colonoscopy at a median of 1 year after EMR. Roughly half of the instances of AN were metachronous lesions. Our data support a 1-year surveillance interval after EMR of LSLs ≥ 10 mm with careful inspection of the entire colon, not just the prior resection site.

The diagnosis of malignant biliary strictures remains problematic, especially in the perihilar region and in primary sclerosing cholangitis (PSC). Conventional cytology obtained during endoscopic retrograde cholangiography (ERC)-guided brushings of biliary strictures is suboptimal due to limited sensitivity, albeit it remains the gold standard with a high specificity. Emerging technologies are being developed and validated to address this pressing unmet patient need. Such technologies include enhanced visualization of the biliary tree by cholangioscopy, intraductal ultrasound, and confocal laser endomicroscopy. Conventional cytology can be aided by employing complementary and advanced cytologic techniques such as fluorescent in situ hybridization (FISH), and this technique should be widely adapted. Interrogation of bile and serum by examining extracellular vesicle number and cargo, and exploiting next-generation sequencing and proteomic technologies, is also being explored. Examination of circulating cell-free deoxyribonucleic acid (cfDNA) for differentially methylated regions is a promising test which is being rigorously validated. The special expertise required for these analyses has to date hampered their validation and adaptation. Herein, we will review these emerging technologies to inform the reader of the progress made and encourage further studies, as well as adaptation of validated approaches.

Self-expanding metal stents (SEMS) are routinely used to palliate malignant dysphagia. However esophageal SEMS can migrate or obstruct due to epithelial hyperplasia. The aim of this study was to evaluate the rates and factors predicting migration and obstruction, and the nutritional outcomes in partially covered (pc) vs. fully covered (fc) SEMS vs. fcSEMS with antimigration fins (AF) placed for malignant dysphagia.A retrospective review of consecutive patients undergoing SEMS placement for malignant dysphagia at three academic medical centers.Among 357 patients, there were 55 (15.4%) stent migrations, 45 (12.6%) obstructions from epithelial hyperplasia, and 20 (5.6%) food impactions. Median overall survival was 79 days (IQR 41,199). The percent weight change/change in albumin at 30 and 60 days after SEMS placement were -2.24%/-0.544 g/dL and -2.98%/-0.55 g/dL, respectively. Stent migration occurred significantly more often with fcSEMS than pcSEMS (25.3% vs 10.9%; P.003), but there was no difference when either group was compared to fcSEMS-AF (19.3%). The overall rate of epithelial hyperplasia resulting in stent obstruction was low (12.6%) and not different between stent types. Factors associated with increased risk of SEMS migration on multivariable logistic regression included stricture traversability with a diagnostic endoscope (OR, 2.37; 95% CI, 1.29-4.35) and use of fcSEMS (OR, 2.56; 1.31-5.00) or fcSEMS-AF (OR, 2.30, 1.03-5.14).Traversability of a malignant esophageal stenosis predicts SEMS migration. In these patients with a limited overall survival, pcSEMS are associated with lower rates of stent migration and similar rates of obstruction compared to fcSEMS.

Demand for anesthesiologist-assisted sedation is expanding for gastrointestinal lower endoscopic procedures and may add to the cost of these procedures. Most lower endoscopy can be accomplished with either no, moderate, or deep sedation; general anesthesia and active airway management are rarely needed. Propofol-based sedation has advantages in terms of satisfaction and recovery over other modalities, but moderate sedation using benzodiazepines and opiates work well for low-risk patients and procedures. No sedation for routine colonoscopy works well for selected patients and eliminates sedation-related risks. There is no difference in outcome measures based on sedation received.

Background and Aims Detection and characterization of focal liver lesions (FLLs) is key for optimizing treatment for patients who may have a primary hepatic cancer or metastatic disease to the liver. This is the first study to develop an EUS-based convolutional neural network (CNN) model for the purpose of identifying and classifying FLLs. Methods A prospective EUS database comprising cases of FLLs visualized and sampled via EUS was reviewed. Relevant still images and videos of liver parenchyma and FLLs were extracted. Patient data were then randomly distributed for the purpose of CNN model training and testing. Once a final model was created, occlusion heatmap analysis was performed to assess the ability of the EUS-CNN model to autonomously identify FLLs. The performance of the EUS-CNN for differentiating benign and malignant FLLs was also analyzed. Results A total of 210,685 unique EUS images from 256 patients were used to train, validate, and test the CNN model. Occlusion heatmap analyses demonstrated that the EUS-CNN model was successful in autonomously locating FLLs in 92.0% of EUS video assets. When evaluating any random still image extracted from videos or physician-captured images, the CNN model was 90% sensitive and 71% specific (area under the receiver operating characteristic [AUROC], 0.861) for classifying malignant FLLs. When evaluating full-length video assets, the EUS-CNN model was 100% sensitive and 80% specific (AUROC, 0.904) for classifying malignant FLLs. Conclusions This study demonstrated the capability of an EUS-CNN model to autonomously identify FLLs and to accurately classify them as either malignant or benign lesions.

Sessile serrated polyps (SSPs) are associated with higher rates of incomplete resection compared to conventional adenomas after traditional snare polypectomy. Outcomes after endoscopic mucosal resection (EMR) are less established. The aim of this study was to evaluate the rate of residual neoplasia at surveillance colonoscopy for SSPs compared to conventional adenomas ≥ 10 mm after en bloc EMR. Retrospective cohort study of consecutive patients referred for EMR of a colonic lesion ≥ 10 mm from 2005 to 2013. Data on procedures, histopathology, and surveillance colonoscopies were recorded. The primary outcome was rate of macroscopically evident residual neoplasia at surveillance colonoscopy for SSPs compared to adenomas. Secondary outcomes included rate of neoplasia at the resection margin. 283 consecutive patients with 293 polyps underwent en bloc EMR including 101 SSPs and 192 adenomas. Pathology commented on the lateral resection margins of the specimen in 235 cases (80%). Of these, neoplasia was noted at the resection margin in 29/64 SSPs (45.3%) compared to 65/171 adenomas (38.0%; P = .37). Surveillance data were available for 153 index lesions with a median interval of 13 months (interquartile range, 10.75–23.25 months). Ten resection sites (6.5%) were found to have residual neoplasia, including 2/52 SSPs (3.8%) and 8/101 adenomas (7.9%; P = .50). Of the cases with surveillance data 128/153 (84%) commented on the lateral margin of the resection specimen. Residual neoplasia was noted in 3/68 lesions (4.4%) with negative margins compared to 5/60 lesions (8.3%) with positive margins (P = .47). En bloc EMR for colonic lesions ≥ 10 mm is associated with a 6.5% rate of macroscopic residual neoplasia. Although 45% of SSPs had neoplasia extending to the resection margin, rates of residual neoplasia at surveillance colonoscopy were low. These results suggest that when feasible en bloc EMR is a reasonable option to resect SSPs ≥ 10 mm.

OBJECTIVES Gastrointestinal bleeding is one of the most common indications for urgent endoscopy in the pediatric setting. The majority of these procedures are performed for control of variceal bleeding, with few performed for nonvariceal upper gastrointestinal (NVUGI) bleeding. The data on therapeutic endoscopy for NVUGI are sparse. The aims of our study were to review our experience with NVUGI bleeding, describe technical aspects and outcomes of therapeutic endoscopy, and determine gastroenterology fellows' training opportunities according to the national training guidelines. METHODS We performed a retrospective review of endoscopy database (Endoworks, Olympus Inc, Center Valley, PA) from January 2009 to December 2014. The search used the following keywords: bleeding, hematemesis, melena, injection, epinephrine, cautery, clip, and argon plasma coagulation. The collected data included demographics, description of bleeding lesion and medical/endoscopic therapy, rate of rebleeding, relevant laboratories, physical examination, and need for transfusion and surgery. The study was approved by the institutional review board. RESULTS During the study period 12,737 upper endoscopies (esophagogastroduodenoscopies) were performed. A total of 15 patients underwent 17 esophagogastroduodenoscopies that required therapeutic intervention to control bleeding (1:750 procedures). The mean ± standard deviation (median) age of patients who required endoscopic intervention was 11.6 ± 6.0 years (14.0 years). Seven out of 17 patients received dual therapy to control the bleeding lesions. All but 3 patients received medical therapy with intravenous proton pump inhibitor, and 3 received octreotide infusions. Six of the patients experienced rebleeding (40%), with 4 out of 6 initially only receiving single modality therapy. Two of these patients eventually required surgical intervention to control bleeding and both patients presented with bleeding duodenal ulcers. There were no cases of aspiration, perforation, or deaths. There were a total of 24 fellows trained in our program during the study period. Less than 1 therapeutic endoscopy per fellow for NVUGI bleeding was performed. CONCLUSIONS NVUGI bleeding requiring therapeutic endoscopic intervention is rare in pediatrics. A high rate (40%) of rebleeding was noted with a large proportion (66%) of patients receiving single modality therapy. Two patients required surgical intervention to control bleeding and both presented with bleeding duodenal ulcers. An insufficient number of therapeutic procedures is available for adequate fellow training requiring supplemental simulator and hands-on animal model, or adult endoscopy unit training.

Background & Aims Biliary complications occur in up to 25% of patient following liver transplantation and are often managed by endoscopic retrograde cholangiopancreatography (ERCP). Pancreatitis is the most common adverse event after ERCP (PEP). Tacrolimus and rectal indomethacin have each been reported to reduce risk of PEP. We investigated the incidence of PEP in patients who have undergone ERCP after liver transplantation and the effectiveness of tacrolimus and/or indomethacin in reducing risk of PEP. Methods We performed a retrospective study of 337 patients who underwent ERCP (n = 937 procedures) for biliary complications after liver transplantation from June 1, 2007 through December 1, 2015. After June 1, 2012, rectal indomethacin (100 mg) was routinely administered at the conclusion of the ERCP unless patients had contraindications. Indomethacin was given after 286 ERCP procedures. After excluding patients with acute/chronic rejection, 323 patients were maintained on a stable dose of tacrolimus prior to ERCP (901 procedures). We collected data on demographic and clinical variables, pre-procedural tacrolimus trough levels, and development of PEP. We calculated adjusted odds ratios (ORs) for the association between tacrolimus and indomethacin use and risk of PEP using mixed-effects multivariable logistic regression. The primary outcome was development of PEP; secondary outcomes included the development moderate-to-severe PEP, cholangitis and bleeding. Results PEP occurred after 2.2% of ERCP procedures. A trough level of tacrolimus above 2.5 ng/mL was associated with 79% lower odds of PEP (OR, 0.21; 95% CI, 0.06–0.72; P = .01). Indomethacin was associated with a 91% reduction in risk of PEP (OR, 0.09; 95% CI, 0.01–0.85; P = .03). Indomethacin use did not affect rates of bleeding or cholangitis or decrease in glomerular filtration rate. In patients with trough levels of tacrolimus above 2.5 ng/mL, addition of indomethacin reduced the odds of PEP by 93% compared with patients who were unexposed to indomethacin. (OR, 0.07; 95% CI, 0.01–0.90; P = .04). Conclusions In a retrospective study of patients who underwent ERCP for biliary complications after liver transplantation, we found trough levels of tacrolimus above 2.5 ng/mL to significantly reduce risk for PEP. Rectal administration of indomethacin after ERCP significantly decreased rates of pancreatitis, and reduced risk further in patients given tacrolimus. Administration of both drugs prevented patients from developing moderate or severe pancreatitis. Indomethacin did not worsen renal function in patients with chronic kidney disease.

Diagnostic tools for nonalcoholic fatty liver disease (NAFLD) detection and prognostication are limited, with histology remaining the criterion standard. We evaluated the feasibility and safety of EUS-guided liver biopsy (EUS-LB) sampling in NAFLD staging.In a prospective cohort of NAFLD patients with steatohepatitis and early liver fibrosis based on magnetic resonance elastography (MRE), EUS-LB sampling procedures were performed using a 22-gauge fork-tip core biopsy needle. Samples were evaluated by a blinded pathologist. Total aggregate sample length (TASL), number of complete portal triads, ability to calculate NAFLD activity score, ability to stage liver fibrosis, and ability to provide enough core liver tissue for lipidomics analysis were evaluated. Performance of EUS-LB sampling was compared with MRE.Forty-one EUS-LB samples were obtained. The median TASL was 2.4 cm (interquartile range, 2.00-2.75). The median number of complete portal triads per TASL was 26 (interquartile range, 7-62). Of the samples, 100% were adequate to convey NAFLD activity score and fibrosis stage. All samples provided enough core liver tissue to allow the application of lipidomics testing. A significant positive linear association between EUS-LB sampling-detected fibrosis and MRE-detected fibrosis was observed (r = .469, P .005). Compared with MRE, EUS-LB sampling established early fibrosis in 13 cases that MRE classified as normal. EUS-LB sampling-related adverse events occurred in 7% and were restricted to postprocedural pain.EUS-LB sampling is a viable technique for full NAFLD evaluation and may be superior to MRE in establishing the diagnosis of nonalcoholic steatohepatitis with early fibrosis. (Clinical trial registration number: NCT02880189.).

Background Select patients with acute cholecystitis (AC) are not candidates for index cholecystectomy. We compared the influence of ERCP-guided transpapillary gallbladder drainage (ERGD) versus percutaneous cholecystostomy (PC) on delayed cholecystectomy outcomes. Methods Consecutive patients undergoing ERGD or PC for AC from January 2007 to October 2018 were included. Primary outcome was the rate of conversion to open cholecystectomy and perioperative complications in groups. Results The study included 52 patients with ERGD and 140 with PC prior to cholecystectomy (median 68 days [IQR: 47–105.5]). Technical success was higher in the PC group (100% vs 91%; P = 0.0004). There was a nonsignificant trend to lower postoperative complications with ERGD (30.7% vs 43.5%; P = 0.07). No difference in conversion to open cholecystectomy OR: 1.5 (95% CI: 0.68–3.65; P = 0.28) or severity of complications (Clavien-Dindo grade >2) OR: 0.60, (95% CI: 0.19–1.87; P = 0.38) was noted between the ERGD and PC groups. PC was associated with higher rates of unplanned repeat intervention (16.4% vs 7.7%; P = 0.02). Conclusion ERGD is suitable for patients with AC who is candidates for delayed cholecystectomy and should be considered for gallbladder drainage in patients with concomitant choledocholithiasis or cholangitis who require ERCP.

Endoscopic retrograde cholangiopancreatography is an effective platform for a variety of therapies in the management of benign and malignant disease of the pancreas. Over the last 50 years, endotherapy has evolved into the first-line therapy in the majority of acute and chronic inflammatory diseases of the pancreas. As this field advances, it is important that gastroenterologists maintain an adequate knowledge of procedure indication, maintain sufficient procedure volume to handle complex pancreatic endotherapy, and understand alternate approaches to pancreatic diseases including medical management, therapy guided by endoscopic ultrasonography, and surgical options.

Ruben D. Acosta, MD, Neena S. Abraham, MD, MSCE, FASGE (invited content expert, ad-hoc member), Vinay Chandrasekhara, MD, Krishnavel V. Chathadi, MD, Dayna S. Early, MD, FASGE, Mohamad A. Eloubeidi, MD, MHS, FASGE, John A. Evans, MD, Ashley L. Faulx, MD, FASGE, Deborah A. Fisher, MD, MHS, FASGE, Lisa Fonkalsrud, BSN, RN, CGRN, Joo Ha Hwang, MD, PhD, FASGE, Mouen A. Khashab, MD, Jenifer R. Lightdale, MD, MPH, FASGE, V. Raman Muthusamy, MD, FASGE, Shabana F. Pasha, MD, John R. Saltzman, MD, FASGE, Aasma Shaukat, MD, MPH, FASGE, Amandeep K. Shergill, MD, Amy Wang, MD, Brooks D. Cash, MD, FASGE, previous Committee Chair, John M. DeWitt, MD, FASGE, Chair