Machine learning for the prediction of post-ERCP pancreatitis risk: A proof-of-concept study.

BACKGROUND: Predicting Post-Endoscopic Retrograde Cholangiopancreatography (ERCP) pancreatitis (PEP) risk can be determinant in reducing its incidence and managing patients appropriately, however studies conducted thus far have identified single-risk factors with standard statistical approaches and limited accuracy. AIM: To build and evaluate performances of machine learning (ML) models to predict PEP probability and identify relevant features. METHODS: A proof-of-concept study was performed on ML application on an international, multicenter, prospective cohort of ERCP patients. Data were split in training and test set, models used were gradient boosting (GB) and logistic regression (LR). A 10-split random cross-validation (CV) was applied on the training set to optimize parameters to obtain the best mean Area Under Curve (AUC). The model was re-trained on the whole training set with the best parameters and applied on test set. Shapley-Additive-exPlanation (SHAP) approach was applied to break down the model and clarify features impact. RESULTS: One thousand one hundred and fifty patients were included, 6.1% developed PEP. GB model outperformed LR with AUC in CV of 0.7 vs 0.585 (p-value=0.012). GB AUC in test was 0.671. Most relevant features for PEP prediction were: bilirubin, age, body mass index, procedure time, previous sphincterotomy, alcohol units/day, cannulation attempts, gender, gallstones, use of Ringer's solution and periprocedural NSAIDs. CONCLUSION: In PEP prediction, GB significantly outperformed LR model and identified new clinical features relevant for the risk, most being pre-procedural.

Predictors of adverse events after endoscopic ultrasound-guided through-the-needle biopsy of pancreatic cysts: a recursive partitioning analysis.

BACKGROUND AND STUDY AIMS : Endoscopic ultrasound-guided through-the-needle biopsy (TTNB) of pancreatic cystic lesions (PCLs) is associated with a non-negligible risk for adverse events (AEs). We aimed to identify the hierarchic interaction among independent predictors for TTNB-related AEs and to generate a prognostic model using recursive partitioning analysis (RPA). PATIENTS AND METHODS : Multicenter retrospective analysis of 506 patients with PCLs who underwent TTNB. RPA of predictors for AEs was performed and the model was validated by means of bootstrap resampling. RESULTS : Mean cysts size was 36.7 mm. Most common diagnoses were intraductal papillary mucinous neoplasm (IPMN, 45 %), serous cystadenoma (18.8 %), and mucinous cystadenoma (12.8 %). Fifty-eight (11.5 %) AEs were observed. At multivariate analysis, age (odds ratio [OR] 1.32, 1.09-2.14; p = 0.05), number of TTNB passes (OR from 2.17, 1.32-4.34 to OR 3.16, 2.03-6.34 with the increase of the number of passes), complete aspiration of the cyst (OR 0.56, 0.31-0.95; p = 0.02), and diagnosis of IPMN (OR 4.16, 2.27-7.69; p < 0.001) were found to be independent predictors of AEs, as confirmed by logistic regression and random forest analyses. RPA identified three risk classes: high-risk (IPMN sampled with multiple microforceps passes, 28 % AEs rate), low-risk (1.4 % AE rate, including patients < 64 years with other-than-IPMN diagnosis sampled with ≤ 2 microforceps passes and with complete aspiration of the cyst) and middle-risk class (6.1 % AEs rate, including the remaining patients). CONCLUSION : TTNB should be selectively used in the evaluation of patients with IPMN. The present model could be applied during patient selection as to optimize the benefit/risk of TTNB.

Transcriptional regulation of chemokine network by biologic monotherapy in ileum of patients with Crohn's disease.

BACKGROUND: Crohn's disease (CD) exacerbation is marked by an intense cellular trafficking. We set out to determine the specific impact of biologic therapies on regulating chemokine network gene expression in healthy, mildly and severely inflamed tissue of CD patients. METHODS: Twenty CD patients on biologics (adalimumab, ustekinumab, vedolizumab) or untreated undergoing colonoscopy due to clinical symptoms of flare. Healthy, mildly and severely inflamed ileum biopsies from each patient were collected. Chemokines and receptors gene expression was analyzed and a STRING analysis for functional enrichment was performed. RESULTS: The chemokine network exhibited wide transcriptional differences among tissues in active untreated patients, whereas all biologic treatments reduced these differences and homogenized their transcriptional activity. In mildly inflamed tissue, all treatments showed gene upregulation while ustekinumab additionally maintained the downregulation of genes such as CCL2, CCL3, CCL17 or CCL23, involved in T cell chemotaxis, inflammatory monocyte and NK trafficking. In severely inflamed tissue, all treatments shared a downregulatory effect on chemokines controlling T cell response (i.e. CXCL16, CXCR3). Adalimumab and vedolizumab significantly reduced the expression of genes promoting antigen presentation by DCs and the initiation of leukocyte extravasation (i.e. CXCL12, CCL25, CCR7). Ustekinumab significantly reduced genes positively regulating Th1 cytokine production and IL-8 mediated signaling (i.e. IL1B, XCL1, CXCR1, CXCR2). CONCLUSION: Biologic therapies differentially target the chemokine network gene expression profile in the ileal tissue of active CD patients. These results may contribute to better understanding cell homing and to defining future personalized therapeutic strategies for CD patients.

Chronic use of statins and acetylsalicylic acid and incidence of post-endoscopic retrograde cholangiopancreatography acute pancreatitis: A multicenter, prospective, cohort study.

OBJECTIVES: Post-endoscopic retrograde cholangiopancreatography (ERCP) acute pancreatitis (PEP) is a frequent complication of this endoscopic procedure. Chronic statin intake has been linked to lower incidence and severity of acute pancreatitis (AP). Periprocedural rectal administration of non-steroidal anti-inflammatory drugs is protective against PEP, but the role of chronic acetylsalicylic acid (ASA) treatment is unclear. We aimed to investigate whether statins and chronic ASA intake are associated with lower risk of PEP. METHODS: An international, multicenter, prospective cohort study. Consecutive patients undergoing ERCP in seven European centers were included. Patients were followed-up to detect those with PEP. Multivariate analysis by means of binary logistic regression was performed, and adjusted odds ratios (aORs) were calculated. RESULTS: A total of 1150 patients were included, and 70 (6.1%) patients developed PEP. Among statins users, 8.1% developed PEP vs. 5.4% among non-users (P = 0.09). Multivariate analysis showed no association between statin use and PEP incidence (aOR 1.68 (95% CI 0.94-2.99, P = 0.08)). Statin use had no effect on severity of PEP, being mild in 92.0% of statin users vs. 82.2% in non-statin users (P = 0.31). Chronic ASA use was not associated with PEP either (aOR 1.02 (95% CI 0.49-2.13), P = 0.96). Abuse of alcohol and previous endoscopic biliary sphincterotomy were protective factors against PEP, while >1 pancreatic guidewire passage, normal bilirubin values, and duration of the procedure >20 minutes, were risk factors. CONCLUSIONS: The use of statins or ASA is not associated with a lower risk or a milder course of PEP.

Endoscopic ultrasound-guided rendezvous in benign biliary or pancreatic disorders with a 22-gauge needle and a 0.018-inch guidewire.

Background and study aims The aim of this study was to assess the efficacy and safety of endoscopic ultrasound-guided rendezvous (EUS-RV) for benign biliary or pancreatic disorders with a 22-gauge needle and a 0.018-inch guidewire. Patients and methods Patients who underwent EUS-RV after failed biliary or pancreatic cannulation for benign disorder were candidates for this study. For EUS-RV, a 22-gauge needle and a 0.018-inch guidewire were used. Inclusion criteria were unsuccessful biliary or pancreatic cannulation for therapeutic endoscopic retrograde cholangiopancreatography (ERCP) with benign biliary or pancreatic obstruction. Exclusion criteria included malignant biliary or pancreatic obstruction, inaccessible papilla due to surgically altered upper gastrointestinal anatomy or duodenal stricture, and previous sphincterotomy and/or biliary stent placement. The primary outcome was the technical success rate of biliary or pancreatic cannulation, and the secondary outcome was the rate of adverse events. Results Thirty-one patients were evaluated. The overall technical success rate was 80.6 % (81.5 % in biliary and 75 % in pancreatic cases). Adverse events (AEs) were identified in 12.9 % of patients, including one with biliary peritonitis, one with abdominal pain and one with severe pancreatitis plus pneumomediastinum. Only one of the AEs (3.3 %) was directly related to the rendezvous procedure. Conclusions EUS-RV may be a safe and feasible salvage method for unsuccessful cannulation for benign disorders. Use of a 22-gauge needle with a 0.018-inch guidewire may be the first option for benign pathology.

Short article: Endoscopic ultrasound-guided fine-needle aspiration of portal vein thrombosis in patients with chronic liver disease and suspicion of hepatocellular carcinoma.

BACKGROUND: Differentiation between benign and malignant portal vein thrombosis (PVT) in the setting of a hepatocellular carcinoma (HCC) is of paramount importance. Histological analysis is usually not carried out because of potential severe side effects of the percutaneous approach. Endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) may be safer and may guide the clinical management of patients with HCC. OBJECTIVE: To describe the feasibility of the EUS-FNA in a series of patients with HCC and PVT. MATERIALS AND METHODS: A chart review of patients with PVT was performed from 2014 to 2016 in three tertiary care hospitals of Spain. Patients with chronic liver disease and PVT with a suspicion of HCC referred for EUS-FNA were included. The impact of the EUS-FNA was assessed by comparing staging following the Barcelona Clinic Liver Cancer algorithm (BCLC) before and after EUS-FNA. RESULTS: Of 104 patients with PVT and chronic liver disease, 23 were considered candidates for EUS-FNA. Eight patients were referred for EUS-FNA. The technique was feasible in seven patients and FNA was positive in six patients. No side effects were reported. EUS-FNA upstaged six out of seven (85.7%) patients: one patient BCLC stage B, two patients BCLC stage A, and three patients in whom the HCC was not diagnosed before EUS-FNA of the PVT. A benign PVT was found in the explant of the only patient with a negative PVT. CONCLUSION: EUS-FNA is a valuable technique in selected patients with chronic liver disease with PVT. It is feasible, safe, and may alter the clinical management in these patients.

Prevalence of incidental pancreatic cyst on upper endoscopic ultrasound.

BACKGROUND: This study aimed to determine the prevalence of incidental pancreatic cysts in patients undergoing upper endoscopic ultrasound without a known pancreatic abnormality. METHODS: This prospective study was conducted in two hospitals in Spain and enrolled consecutive patients referred for upper endoscopic ultrasound for a condition unrelated to the pancreas. Patients with a previous pancreatic anomaly, history of acute or chronic pancreatitis, evidence of acute pancreatitis, previous upper gastrointestinal surgery, or chronic abdominal pain suggestive of pancreatic origin were excluded. Univariate logistic regression was performed to evaluate individual covariates and the incidental pancreatic cyst risk. RESULTS: A total of 298 patients were included, of whom 64 had pancreatic cysts (21.5%; 16.9-26.6%). The mean size of the cysts was 6.3±3.7 (range 3-25) mm. Six cysts (2%) were >10 mm and 16 (5.4%) were compatible with branch duct intraductal papillary mucinous neoplasm. The pancreatic cyst prevalence was similar in the two hospitals and increased significantly with age. CONCLUSION: The prevalence of incidental pancreatic cysts during endoscopic ultrasound was very high in our study population.

KRAS and BRAF somatic mutations in colonic polyps and the risk of metachronous neoplasia.

BACKGROUND & AIMS: High-risk features of colonic polyps are based on size, number, and pathologic characteristics. Surveillance colonoscopy is often recommended according to these findings. This study aimed to determine whether the molecular characteristics of polyps might provide information about the risk of metachronous advanced neoplasia. METHODOLOGY: We retrospectively included 308 patients with colonic polyps. A total of 995 polyps were collected and tested for somatic BRAF and KRAS mutations. Patients were classified into 3 subgroups, based on the polyp mutational profile at baseline, as follows: non-mutated polyps (Wild-type), at least one BRAF-mutated polyp, or at least one KRAS-mutated polyp. At surveillance, advanced adenomas were defined as adenomas ≥ 10 mm and/or with high grade dysplasia or a villous component. In contrast, advanced serrated polyps were defined as serrated polyps ≥ 10 mm in any location, located proximal to the splenic flexure with any size or with dysplasia. RESULTS: At baseline, 289 patients could be classified as wild-type (62.3%), BRAF mutated (14.9%), or KRAS mutated (22.8%). In the univariate analysis, KRAS mutations were associated with the development of metachronous advanced polyps (OR: 2.36, 95% CI: 1.22-4.58; P = 0.011), and specifically, advanced adenomas (OR: 2.42, 95% CI: 1.13-5.21; P = 0.023). The multivariate analysis, adjusted for age and sex, also showed associations with the development of metachronous advanced polyps (OR: 2.27, 95% CI: 1.15-4.46) and advanced adenomas (OR: 2.23, 95% CI: 1.02-4.85). CONCLUSIONS: Our results suggested that somatic KRAS mutations in polyps represent a potential molecular marker for the risk of developing advanced neoplasia.

Initial experience with EUS-guided cholangiopancreatography for biliary and pancreatic duct drainage: a Spanish national survey.

BACKGROUND: EUS-guided cholangiopancreatography (ESCP) allows transmural access to biliopancreatic ducts when ERCP fails. Data regarding technical details, safety, and outcomes of ESCP are still unknown. OBJECTIVE: To evaluate outcomes of ESCP in community and referral centers at the initial development phase of this procedure, to identify the ESCP stages with higher risk of failure, and to evaluate the influence on outcomes of factors related to the endoscopist. DESIGN: Multicenter retrospective study. SETTING: Public health system hospitals with experience in ESCP in Spain. PATIENTS: A total of 125 patients underwent ESCP in 19 hospitals, with an experience of <20 procedures. INTERVENTION: ESCP. MAIN OUTCOME MEASUREMENTS: Technical success and complication rates in the initial phase of implantation of ESCP are described. The influence of technical characteristics and endoscopist features on outcomes was analyzed. RESULTS: A total of 125 patients from 19 hospitals were included. Biliary ESCP was performed in 106 patients and pancreatic ESCP was performed in 19. Technical success was achieved in 84 patients (67.2%) followed by clinical success in 79 (63.2%). Complications occurred in 29 patients (23.2%). Unsuccessful manipulation of the guidewire was responsible for 68.2% of technical failures, and 58.6% of complications were related to problems with the transmural fistula. LIMITATIONS: Retrospective study. CONCLUSION: Outcomes of ESCP during its implantation stage reached a technical success rate of 67.2%, with a complication rate of 23.2%. Intraductal manipulation of the guidewire seems to be the most difficult stage of the procedure.

Direct intracystic biopsy and pancreatic cystoscopy through a 19-gauge needle EUS (with videos).

BACKGROUND: Histologic diagnosis of cystic pancreatic lesions (CPLs) is often difficult because of the low sensitivity of FNA and brush cytology. OBJECTIVE: To discover whether obtaining biopsy samples from the cystic wall could increase the diagnostic yield of these lesions. DESIGN: A pilot study including 2 patients with CPLs. SETTING: Endoscopy unit in a tertiary-care hospital. PATIENTS: Two women with CPLs located at the pancreatic head. INTERVENTIONS: On EUS, biopsy forceps and a SpyGlass fiberoptic were passed through a 19-gauge needle to visualize and obtain samples from the cystic wall. MAIN OUTCOME MEASUREMENTS: The histologic assessment was based on the obtained biopsy samples. RESULTS: Both CPLs were considered to be mucinous cystoadenomas, because mucinous-like cylindric epithelium without cellular atypia was observed. LIMITATIONS: Pilot study. CONCLUSIONS: Obtaining biopsy samples from the wall of a CPL is now feasible. It represents a significant advantage in the diagnostic yield of this type of lesion.

Deep sedation with propofol does not precipitate hepatic encephalopathy during elective upper endoscopy.

BACKGROUND: The risk of hepatic encephalopathy (HE) precipitated by propofol has not been established. OBJECTIVE: To know whether the use of propofol for endoscopy in patients with cirrhosis induces minimal or overt HE. DESIGN: A cohort study. SETTING: A tertiary-referral university hospital endoscopy unit. PATIENTS: Patients with cirrhosis who received an endoscopy with propofol sedation. We excluded patients with clinical HE. A group of patients without liver disease was included to compare the incidence of adverse effects. INTERVENTIONS: Minimal HE (MHE) was diagnosed by using the psychometric hepatic encephalopathy score (PHES) battery of psychometric tests. Cognitive status before and 1 hour after the endoscopy was evaluated by measuring the critical flicker frequency (CFF). MAIN OUTCOME MEASUREMENTS: Overt and minimal HE. RESULTS: Twenty patients with cirrhosis and 20 patients without cirrhosis were included. There were no differences between groups in the incidence of adverse effects. Thirteen patients (65%) had MHE before the endoscopy. No patient developed overt HE after sedation. We did not observe differences in CFF before and after sedation in patients without MHE: median (25th-75th percentile), 40.8 Hz (37.1-46.0 Hz) versus 41.1 Hz (36.0-44.3 Hz), P = .8). None of the patients who were without MHE showed a decrease in the CFF under the cutoff of 39 Hz after sedation. There were not significant changes in CFF before and after propofol sedation in patients previously diagnosed of MHE: median (25th-75th percentile), 40.6 Hz (36.8-49.1 Hz) versus 42.7 Hz (36.8-52.4 Hz), P = .08. LIMITATION: A small number of patients were included in the study. CONCLUSIONS: The use of propofol in the sedation of patients with cirrhosis during endoscopic procedures does not precipitate minimal or overt HE.

Right lateral position does not affect gastric transit times of video capsule endoscopy: a prospective study.

BACKGROUND: Video capsule endoscopy (VCE) examination of the small bowel is not complete in approximately 20% of the procedures. This fact limits its diagnostic yield. One of the main factors that influences the small-bowel transit time (SBTT) is the gastric transit time (GTT), ie, the interval in which the capsule stays in the stomach. It has been described that placing the patient in a right lateral position (RLP) after swallowing the capsule could decrease the GTT. OBJECTIVE: To investigate whether the RLP, after the patient swallows the capsule, shortens the GTT and, secondarily, increases the rate of complete procedures. DESIGN: Randomized prospective study. SETTING: Third-level hospital. PATIENTS: Consecutive outpatients in whom VCE was indicated. Exclusion criteria were inpatients and previous gastric surgery. INTERVENTION: GTT for RLP 30 minutes after swallowing the capsule versus non-RLP (standing up position). MAIN OUTCOME MEASUREMENTS: The GTT, SBTT, and rate of complete procedures (examination of the entire small bowel). RESULTS: We did not observe significant differences in the GTT, the SBTT, and the complete procedures between groups. LIMITATION: Only outpatients were included. CONCLUSIONS: RLP after swallowing the capsule does not influence either GTT nor the rate of VCE complete procedures.