Risk factors for EUS-guided radiofrequency ablation adverse events in patients with pancreatic neoplasms: a large national French study (RAFPAN study).

BACKGROUND AND AIMS: EUS-guided radiofrequency ablation (EUS-RFA) has been described as a potentially curative option for solid and cystic pancreatic neoplasms. We aimed to assess the safety and efficacy of pancreatic EUS-RFA in a large study population. METHODS: A retrospective study retrieving all consecutive patients who underwent pancreatic EUS-RFA during 2019 and 2020 in France was conducted. Indication, procedural characteristics, early and late adverse events (AEs), and clinical outcomes were recorded. Risk factors for AEs and factors related to complete tumor ablation were assessed on univariate and multivariate analyses. RESULTS: One hundred patients (54% men, 64.8 ± 17.6 years old) affected by 104 neoplasms were included. Sixty-four neoplasms were neuroendocrine neoplasms (NENs), 23 were metastases, and 10 were intraductal papillary mucinous neoplasms with mural nodules. No procedure-related mortality was observed, and 22 AEs were reported. Proximity of pancreatic neoplasms (≤1 mm) to the main pancreatic duct was the only independent risk factor for AEs (odds ratio [OR), 4.10; 95% confidence interval [CI), 1.02-15.22; P = .04). Fifty-nine patients (60.2%) achieved a complete tumor response, 31 (31.6%) a partial response, and 9 (9.2%) achieved no response. On multivariate analysis, NENs (OR, 7.95; 95% CI, 1.66-51.79; P < .001) and neoplasm size <20 mm (OR, 5.26; 95% CI, 2.17-14.29; P < .001) were independently related to complete tumor ablation. CONCLUSIONS: The results of this large study confirm an overall acceptable safety profile for pancreatic EUS-RFA. Close proximity (≤1 mm) to the main pancreatic duct represents an independent risk factor for AEs. Good clinical outcomes in terms of tumor ablation were observed, especially for small NENs.

Endoscopic training during fellowship: A nationwide French study: Impact of theoretical courses and simulation-based training during fellowship.

BACKGROUND: In France, it is mandatory that gastroenterology fellows have mastered the basic level of endoscopy by the end of training. The aim of this study was to assess improvement in the quality of fellows' endoscopy training in France during the last four years. METHODS: All fellows in France in training were eligible for participation. A 21-item questionnaire was sent out. The primary outcome was the completion by fourth year fellows of all the number of procedures recommended. Results were compared with those of a 2016 survey. RESULTS: Two-hundred-and-sixty-five fellows responded to the survey. The participation rate was 47.0%. The mean age was 27.3 ± 1.0 years and 56.4% were female. Access to theoretical courses (63.7% vs. 30.6%, p < 0.001) and simulation-based training (virtual reality simulator: 58.4% vs. 28.2%, p < 0.001, animal models: 29.4% vs. 17.2%, p < 0.001) was significantly higher in 2020. Although the number of procedures did not increase, significantly higher perception of skill acquisition in colonoscopy as well as diminished pressure to advance procedures were noted. CONCLUSION: Access to theoretical courses and simulation-based training and perceived acquisition of numerous skills has gotten better. However, the quality of training in endoscopy still needs improvement.

Endoscopic balloon dilation of colorectal strictures complicating Crohn's disease: a multicenter study.

INTRODUCTION: While endoscopic balloon dilation (EBD) is widely used to manage ileal strictures, EBD of colorectal strictures remains poorly investigated in Crohn's disease (CD). METHODS: We performed a retrospective study that included all consecutive CD patients who underwent EBD for native or anastomotic colorectal strictures in 9 tertiary centers between 1999 and 2018. Factors associated with EBD failure were also investigated by logistic regression. RESULTS: Fifty-seven patients (25 women, median age: 36 years (InterQuartile Range, 31-48) were included. Among the 60 strictures, 52 (87%) were native, 39 (65%) measured < 5 cm and the most frequent location was the left colon (27%). Fifty-seven (95%) were non-passable by the scope and 35 (58%) were ulcerated. Among the 161 EBDs performed (median number of dilations per stricture: 2, IQR 1-3), technical and clinical success were achieved for 79% (n = 116/147) and 77% (n = 88/115), respectively. One perforation occurred (0.6% per EDB and 2% per patient). After a median follow-up of 4.3 years (IQR 2.0-8.4), 24 patients (42%) underwent colonic resection and 24 (42%) were asymptomatic without surgery. One colon lymphoma and one colorectal cancer were diagnosed (3.5% of patients) from endoscopic biopsies and at the time of surgery, respectively. No factor was associated with technical or clinical success. CONCLUSION: EDB of CD-associated colorectal strictures is feasible, efficient and safe, with more than 40% becoming asymptomatic without surgery.

Safety of Endoscopic Ultrasound-Guided Fine-Needle Aspiration for Pancreatic Solid Pseudopapillary Neoplasm Before Surgical Resection: A European Multicenter Registry-Based Study on 149 Patients.

OBJECTIVES: The results of only a few endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) for pancreatic solid pseudopapillary neoplasm (SPN) have been published, and the safety of the procedure has never been investigated. Our study compared the recurrence rate in patients with and without preoperative EUS-FNA. METHODS: This European multicenter registry-based study was conducted in 22 digestive units, and retrospectively included all patients who underwent complete resection of a pancreatic SPN from 2000 to 2018. Patients with and without initial EUS-FNA were compared, and postsurgery recurrence and the associated risk factors were evaluated. RESULTS: A complete resection of a pancreatic SPN was performed in 149 patients (133 women, 89%), with a mean age of 34 (standard deviation, 14) years. There were no significant differences between the with (78 patients) and without (71 patients) EUS-FNA groups, except for age and tumor size and location.Preoperative EUS-FNA allowed pancreatic SPN diagnosis in 63/78 cases (81%). After a mean follow-up of 43 (standard deviation, 36) months, recurrence was noted in 4 patients (2.7%). Preoperative EUS-FNA was not correlated with recurrence, but an older age (P = 0.005) was significant. CONCLUSIONS: Preoperative EUS-FNA does not affect pancreatic SPN recurrence. In this series, old age was significantly correlated with recurrence.

A simplified table using validated diagnostic criteria is effective to improve characterization of colorectal polyps: the CONECCT teaching program.

Introduction and study aims Accurate real-time endoscopic characterization of colorectal polyps is key to choosing the most appropriate treatment. Mastering the currently available classifications is challenging. We used validated criteria for these classifications to create a single table, named CONECCT, and evaluated the impact of a teaching program based on this tool. Methods A prospective multicenter study involving GI fellows and attending physicians was conducted. During the first session, each trainee completed a pretest consisting in histological prediction and choice of treatment of 20 colorectal polyps still frames. This was followed by a 30-minute course on the CONECCT table, before taking a post-test using the same still frames reshuffled. During a second session at 3 - 6 months, a last test (T3 M) was performed, including these same still frames and 20 new ones. Results A total 419 participants followed the teaching program between April 2017 and April 2018. The mean proportion of correctly predicted/treated lesions improved significantly from pretest to post-test and to T3 M, from 51.0 % to 74.0 % and to 66.6 % respectively ( P  < 0.001). Between pretest and post-test, 343 (86.6 %) trainees improved, and 153 (75.4 %) at T3 M. Significant improvement occurred for each subtype of polyp for fellows and attending physicians. Between the two sessions, trainees continued to progress in the histology prediction and treatment choice of polyps CONECCT IIA. Over-treatment decreased significantly from 30.1 % to 15.5 % at post-test and to 18.5 % at T3 M ( P  < 0.001). Conclusion The CONECCT teaching program is effective to improve the histology prediction and the treatment choice by gastroenterologists, for each subtype of colorectal polyp.

Interventional endoscopic ultrasound: A new promising way for intrahepatic portosystemic shunt with portal pressure gradient.

BACKGROUND AND OBJECTIVES: Interventional endoscopic ultrasound (EUS) is a promising novel approach for intravascular interventions. The aim of this study was to assess the feasibility and safety of a EUS-guided intrahepatic portosystemic shunt (EGIPS) with portal pressure gradient measurement in a live porcine model. METHODS: The left hepatic vein (LHV) or the inferior vena cava (IVC) was punctured with a needle that advanced into the portal vein (PV). A guidewire was then inserted into the PV, and a needle knife was used to create an intrahepatic fistula between LHV and PV. Portal pressure was recorded. The fistula was dilated with a balloon and a biliary metal stent was deployed between LHV and PV under sonographic and fluoroscopic observation. A portocavography validated the patency of the stent. Necropsies were realized after euthanasia. RESULTS: Portosystemic stenting was achieved in 19/21 pigs. Final portocavography confirmed stent patency between PV and LHV or IVC in 17 pigs (efficacy of 81%): Four stents were dysfunctional as two were thrombosed and two were poor positioned. Portal pressure was documented before and after shunting in 20/21 pigs. Necropsies revealed that 19/21 procedures were transesophageal and two were transgastric. Hemoperitoneum and pneumothorax were found in one pig and hemothorax was found in two pigs. Morbidity was 14.2% (3/21 animals). CONCLUSION: EGIPS was feasible in 91% of cases, functional in 81%, with 14.2% per procedure morbidity. EGIPS still needs to be assessed in portal hypertension pig models with longer follow-up before being considered as an alternative when the transjugular intrahepatic portosystemic shunt fails.

Usefulness of the Hook knife in flexible endoscopic myotomy for Zenker's diverticulum.

AIM: To investigate the outcome of flexible endoscopic myotomy performed with the Hook knife in patients with symptomatic Zenker's diverticulum (ZD). METHODS: All consecutive patients treated for ZD at our institution between 7/2012 and 12/2016 were included. The flexible endoscopic soft diverticuloscope-assisted technique with endoclips placement and Hook knife myotomy were performed in all patients. Here we report a retrospective review of prospectively collected data. Demographics, dysphagia score (Dakkak and Bennett), associated symptoms and adverse events were collected pre-procedure, at 2 and 6 mo post-procedure, and at the end of the follow-up period. Clinical success was defined as at least 1-point improvement in dysphagia score and a residual dysphagia score ≤ 1, with no need for reintervention. Dysphagia scores were compared before treatment and at end-of-follow-up using the Wilcoxon test. RESULTS: Twenty-four patients were included. Mean size of ZD was 3.0 cm (range 2-8 cm). Mean number of sessions was 1.17/patient (range 1-3 sessions). Overall clinical success was 91.7%. Two adverse events (8.3%) occurred, and both were managed conservatively. No bleeding or perforation was reported. Mild pain was reported by 9 patients (37.5%). Median hospital stay was 1 d (range 1-6). Median follow-up was 19.5 mo (range 6-53). Mean ± SD dysphagia score was 2.25 ± 0.89 before treatment and decreased to 0.41 ± 0.92 at end-of-follow-up (P < 0.001). Regurgitation and cough dropped from 91.7% and 50% to 12.5% and 0% at the end of follow-up, respectively. Recurrence was observed in 3 patients, and all 3 were symptom-free after one more session. CONCLUSION: The Hook knife, used in the soft diverticuloscope-assisted technique setting, is efficient and safe for treatment of ZD.

Complete endoscopic sphincterotomy with vs. without large-balloon dilation for the removal of large bile duct stones: randomized multicenter study.

Background and study aims Endoscopic sphincterotomy plus large-balloon dilation (ES-LBD) has been reported as an alternative to endoscopic sphincterotomy for the removal of bile duct stones. This multicenter study compared complete endoscopic sphincterotomy with vs. without large-balloon dilation for the removal of large bile duct stones. This is the first randomized multicenter study to evaluate these procedures in patients with exclusively large common bile duct (CBD) stones. Methods Between 2010 and 2015, 150 patients with one or more common bile duct stones ≥ 13 mm were randomized to two groups: 73 without balloon dilation (conventional group), 77 with balloon dilation (ES-LBD group). Mechanical lithotripsy was subsequently performed only if the stones were too large for removal through the papilla. Endoscopic sphincterotomy was complete in both groups. Patients could switch to ES-LBD if the conventional procedure failed. Results There was no between-group difference in number and size of stones. CBD stone clearance was achieved in 74.0 % of patients in the conventional group and 96.1 % of patients in the ES-LBD group (P < 0.001). Mechanical lithotripsy was needed significantly more often in the conventional group (35.6 % vs. 3.9 %; P < 0.001). There was no difference in terms of morbidity (9.3 % in the conventional group vs. 8.1 % in the ES-LBD group; P = 0.82). The cost and procedure time were not significantly different between the groups overall, but became significantly higher for patients in the conventional group who underwent mechanical lithotripsy. The conventional procedure failed in 19 patients, 15 of whom underwent a rescue ES-LBD procedure that successfully cleared all stones. Conclusions Complete endoscopic sphincterotomy with large-balloon dilation for the removal of large CBD stones has similar safety but superior efficiency to conventional treatment, and should be considered as the first-line step in the treatment of large bile duct stones and in rescue treatment.Trial registered at ClinicalTrials.gov (NCT02592811).

Adenoma detection with blue-water infusion colonoscopy: a randomized trial.

Background and aims Colonoscopy is currently the reference method to detect colorectal neoplasia, yet some adenomas remain undetected. The water infusion technique and dying with indigo carmine has shown interesting results for reducing this miss rate. The aim of this study was to compare the adenoma detection rate (adenoma and adenocarcinoma; ADR) and the mean number of adenomas per patient (MAP) for blue-water infusion colonoscopy (BWIC) versus standard colonoscopy. Methods We performed a multicenter, randomized controlled trial in eight units, including patients with a validated indication for colonoscopy (symptoms, familial or personal history, fecal occult blood test positive). Consenting patients were randomized 1:1 to BWIC or standard colonoscopy. All colonoscopies were performed by experienced colonoscopists. All colonoscopy quality indicators were prospectively recorded. Results Among the 1065 patients included, colonoscopies were performed completely for 983 patients (514 men; mean age 59.1). The ADR was not significantly different between the groups; 40.4 % in the BWIC group versus 37.5 % in the standard colonoscopy group (odds ratio [OR] 1.13; 95 % confidence interval [CI] 0.87 - 1.48; P = 0.35). MAP was significantly greater in the BWIC group (0.79) than in the standard colonoscopy group (0.64; P = 0.005). For advanced adenomas, the results were 50 (10.2 %) and 36 (7.3 %), respectively (P = 0.10). The cecal intubation rate was not different but the time to cecal intubation was significantly longer in BWIC group (9.9 versus 6.2 minutes; P < 0.001). Conclusion Despite the higher MAP with BWIC, the routine use of BWIC does not translate to a higher ADR. Whether increased detection ultimately results in a lower rate of interval carcinoma is not yet known. CLINICAL TRIALS REGISTRATION: EudraCT 2012-A00548 - 35; NCT01937429.

Long-term outcome of endoscopic ultrasound-guided pelvic abscess drainage: a two-center series.

Background and study aim Endoscopic ultrasound (EUS)-guided pelvic abscess drainage has been reported but long-term data remain limited. This two-center study evaluated long-term outcome of EUS-guided pelvic abscess drainage. Patients and methods Between May 2003 and December 2015, 37 consecutive patients were treated for perirectal or perisigmoid abscesses via EUS-guided drainage using plastic or lumen-apposing metal stent (LAMS). Clinical success was defined as complete resolution of the abscess on follow-up computed tomography (CT) scan at 4 weeks with symptom relief. Long-term success was defined as abscess resolution without the need for surgery and without recurrence on long-term follow-up (> 12 months). Results Median abscess size was 60 mm (interquartile range 41 - 70). Causes were postsurgical (n = 31, 83.8 %) or secondary to medical conditions (n = 6, 16.2 %). EUS-guided drainage involved needle aspiration (n = 4), plastic stent placement (n = 29) or LAMS placement (n = 4 patients). Technical and clinical success was achieved in 37 patients (100 %; 95 % confidence interval [CI] 91 - 100) and 34 patients (91.9 %; 95 %CI 78 - 98), respectively (5 patients needed a second EUS-guided intervention within 14 days after drainage). One patient required surgery and one required best supportive care owing to persistent abscess. Early complications were perforation requiring surgery (n = 1), stent migration (n = 1), and rectal discomfort (n = 1). At a median follow-up of 64 months (IQR 19 - 81), two patients experienced abscess recurrence, at 3 and 12 months, respectively, and were treated surgically. Long-term success was achieved in 32 of 37 patients (86.5 %; 95 %CI 71 - 95). Conclusion EUS-guided drainage of pelvic abscess is safe, has good long-term outcome, and should be considered as an alternative to percutaneous and surgical drainage.

Temporary placement of a covered duodenal stent can avoid riskier anterograde biliary drainage when ERCP for obstructive jaundice fails due to duodenal invasion.

BACKGROUND: Duodenal stenosis is one of the most common causes of failed ERCP for obstructive jaundice. Alternative approaches include anterograde biliary drainage, with higher morbidity. We report in this study the efficacy and safety of temporary placement of a covered duodenal self-expandable metal stent (cSEMS) in order to access the papilla and achieve secondary retrograde biliary drainage in patients with obstructive jaundice and failed ERCP due to concomitant duodenal stenosis. METHODS: From June 2006 to March 2014, a total of 26 consecutive patients presenting obstructive jaundice without severe sepsis with failed ERCP due to duodenal invasion were enrolled. A temporary 7-day duodenal cSEMS was placed during the failed ERCP, and a second ERCP was attempted at day 7 after duodenal stent removal. RESULTS: Duodenal cSEMS placement and retrieval were technically successful in all cases. Access to the papilla at day 7 was possible in 25 cases (96 %, 95 % CI 80-99 %). Secondary successful ERCP was achieved in 19 cases (76 %, 95 % CI 55-91 %, i.e., 73 %, 95 % CI 73-86 %, in an intention-to-treat analysis). Mean bilirubin level was 102 ± 90 µmol/L at baseline rising to 164 ± 121 µmol/L at day 7. There were 6 stent migrations and no adverse events recorded between the two ERCPs. CONCLUSIONS: When ERCP for obstructive jaundice fails due to duodenal invasion, temporary cSEMS placement offers a safe and effective way to achieve successful secondary ERCP while avoiding riskier endoscopic ultrasound or percutaneous transhepatic anterograde biliary drainage.

Endoscopic treatment of benign esophageal strictures: a literature review.

INTRODUCTION: Benign esophageal strictures arise from various etiologies and are frequently encountered. Although endoscopic dilation is still the first-line therapy, recurrent strictures do occur in approximately 10% of the cases and remains a challenge to gastroenterologists. Areas covered: A literature search was performed using PubMed and Google Scholar databases for original and review articles on endoscopic treatment of benign esophageal strictures. This review outlines the main available treatment options and its controversies in the management of refractory benign esophageal strictures. Expert commentary: Adding local steroid injections to dilation can be effective for peptic stenosis and strictures after endoscopic submucosal dissection, but remains uncertain for anastomotic strictures. Intralesional injections of mitomycin-C could be useful in corrosive strictures. Incisional therapy can be a reliable alternative in Schatzki rings and in anastomotic strictures, in experienced hands. By contrast, long-term outcome with endoprosthetic treatment is disappointing, and stent placement should be carefully considered and individualized.

Computer-Aided Classification of Gastrointestinal Lesions in Regular Colonoscopy.

We have developed a technique to study how good computers can be at diagnosing gastrointestinal lesions from regular (white light and narrow banded) colonoscopic videos compared to two levels of clinical knowledge (expert and beginner). Our technique includes a novel tissue classification approach which may save clinician's time by avoiding chromoendoscopy, a time-consuming staining procedure using indigo carmine. Our technique also discriminates the severity of individual lesions in patients with many polyps, so that the gastroenterologist can directly focus on those requiring polypectomy. Technically, we have designed and developed a framework combining machine learning and computer vision algorithms, which performs a virtual biopsy of hyperplastic lesions, serrated adenomas and adenomas. Serrated adenomas are very difficult to classify due to their mixed/hybrid nature and recent studies indicate that they can lead to colorectal cancer through the alternate serrated pathway. Our approach is the first step to avoid systematic biopsy for suspected hyperplastic tissues. We also propose a database of colonoscopic videos showing gastrointestinal lesions with ground truth collected from both expert image inspection and histology. We not only compare our system with the expert predictions, but we also study if the use of 3D shape features improves classification accuracy, and compare our technique's performance with three competitor methods.

The Clinical Outcome in Patients Treated With a Newly Designed SEMS in Cervical Esophageal Strictures and Fistulas.

BACKGROUND AND GOALS: Using a self-expandable metallic stent (SEMS) in the cervical esophagus is controversial due to an increased risk of complications. Here we assessed a new type of SEMS purpose-designed for the cervical esophagus area. STUDY: Patients with malignant or benign stenosis within 4 cm distance of the upper esophageal sphincter who underwent placement of a SEMS with a shorter proximal head (Niti-S Esophageal Covered Stent-Cervical-type, NSECSC), were included. Main outcome measures were the functional outcome, tolerance, complications, recurrent dysphagia, and survival. RESULTS: About 37 patients had an NSECSC placed between April 2008 and June 2013 for esophageal stenosis (malignant=20, benign=17), 5 with associated tracheoesophageal fistula. The mean stenosis-upper esophageal sphincter distance was 1.86±1.27 cm. The median follow-up was 150 days. Dysphagia improved in 27/37 cases (73%). Short-term and long-term tolerance without needing stent removal was 92% and 82%, respectively. The complication rate was 59% (22/37): 32% (n=14) major complications [fistula (3), perforation (3), aspiration pneumonia (5), laryngeal dyspnea (2), and bleeding (1)], and 27% (n=10) minor complications [pain (7) or dysphonia (3)]. A multivariate analysis confirmed a higher risk of major complications in cases of benign stenosis (odds ratio=5.2; 95% confidence interval, 1.05-25.90; P=0.04). Recurrent dysphagia occurred in 15 patients (obstruction=7, migration=8). CONCLUSIONS: The NSECSC does not appear less morbid than standard SEMS in the cervical esophageal area, but could be useful in malignant indications as it is well-tolerated and offers effective palliation of the dysphagia. However, this device should not be used in benign cervical esophageal strictures or fistulas.

Usefulness of EUS-guided fine needle aspiration biopsy in the diagnosis of suspected or recurring lymphoproliferative disorders.

BACKGROUND: EUS-guided fine needle aspiration biopsy (EUS-FNAB) of deep-seated lymphadenopathy is proposed to identify lymphoproliferative disorders when no superficial lesion is accessible. METHODS: We analyzed prospectively collected data of 115 EUS-FNABs from 73 thoracic or abdomino-pelvic targets in 52 patients with suspected lymphoproliferative disorders (LPDs) between January 2005 and May 2011 from a single institution. Conventional histology and immunohistochemistry procedures were performed on samples. RESULTS: No complications were recorded. An LPD was identified in 29 cases and ruled out in 21 cases. In 2 cases the analysis was negative, but an LPD was identified using a secondary procedure. For the identification of LPDs irrespective of subtype, this procedure has positive and negative predictive values of 100% and 91.3% respectively, with 93.6% sensitivity and 100% specificity. In 31 patients finally diagnosed with LPDs, an accurate diagnosis meeting the 2008 World Health Organization classification criteria was established in 21 (68%) cases, success being significantly associated with target size above 30 mm in multivariate analysis (odds ratio 7.47; p = 0.05). CONCLUSION: EUS-FNAB of deep-seated lymphadenopathy with conventional morphological assessment appears to have a high diagnostic value for LPD identification and can obviate invasive surgery. A sub-classification was possible in two thirds of the cases.

Endoscopic ultrasound-guided biliary drainage after failed ERCP: cumulative experience of 101 procedures at a single center.

BACKGROUND AND STUDY AIM: Endoscopic ultrasound (EUS)-guided biliary access is an alternative to percutaneous access after failed endoscopic retrograde cholangiopancreatography (ERCP). This report presents 7 years' cumulative experience of EUS-guided biliary drainage for obstructive jaundice in patients with failed ERCP. PATIENTS AND METHODS: Between February 2006 and February 2013, 101 patients (malignant = 98, benign = 3) with previous failed ERCP underwent an EUS intra- or extrahepatic approach with transluminal stenting or an EUS-guided rendezvous procedure with transpapillary stent placement. A single endoscopist performed all procedures. RESULTS: A total of 71 patients underwent the intrahepatic approach (66 hepatogastrostomies and 5 EUS-guided rendezvous), and 30 underwent the extrahepatic approach (26 choledochoduodenostomies, 1 choledochojejunostomy, 1 choledochoantrostomy, and 2 EUS-guided cholangiographies). Technical and clinical success rates were 98.0 % and 92.1 %, respectively. There was no difference in efficacy between hepatogastrostomies and choledochoduodenostomies (94 % vs. 90 %; P = 0.69) or in major complications (10.6 % vs. 6.7 %; P = 1). Adverse events occurred in 12 patients (11.9 %): 10 in the hepatogastrostomy group (2 limited pneumoperitoneum, 1 hepatic hematoma, 5 bile leakage, 2 sepsis), and 2 in the choledochoduodenostomy group (1 arteriobiliary fistula and 1 sepsis). There were six procedure-related deaths, five among the first 50 patients and one among the last 51 patients. Hepatogastrostomy vs. choledochoduodenostomy, plastic vs. metal stenting, stent-in-stent vs. 1 stent, nasobiliary drain, or postoperative octreotide infusion were not prognostic of bile leakage. CONCLUSION: EUS-guided biliary drainage is an efficient technique, but is associated with significant morbidity that seems to decrease with the learning curve. It should be performed in tertiary care centers in selected patients. Prospective randomized studies are needed to compare EUS-guided biliary drainage with percutaneous transhepatic cholangiography drainage.