Development and Validation of a Machine Learning-based, Point-of-Care Risk Calculator for Post-ERCP Pancreatitis and Prophylaxis Selection.

BACKGROUND AND AIMS: A robust model of post-ERCP pancreatitis (PEP) risk is not currently available. We aimed to develop a machine learning-based tool for PEP risk prediction to aid in clinical decision-making related to periprocedural prophylaxis selection and post-procedural monitoring. METHODS: Feature selection, model training, and validation were performed using patient-level data from 12 randomized controlled trials. A gradient-boosted machine (GBM) model was trained to estimate PEP risk and the performance of the resulting model was evaluated using the area under the receiver operating curve (AUC) with 5-fold cross-validation. A web-based clinical decision-making tool was created, and a prospective pilot study was performed using data from ERCPs performed at the Johns Hopkins Hospital over a one-month period. RESULTS: A total of 7389 patients were included in the GBM with an 8.6% rate of PEP. The model was trained on twenty PEP risk factors and 5 prophylactic interventions (rectal non-steroidal anti-inflammatory drugs [NSAID], aggressive hydration, combined rectal NSAID and aggressive hydration, pancreatic duct [PD] stenting, and combined rectal NSAID and PD stenting). The resulting GBM model had an AUC of 0.70 (65% specificity, 65% sensitivity, 95% negative predictive value, 15% positive predictive value). A total of 135 patients were included in the prospective pilot study, resulting in an AUC of 0.74. CONCLUSIONS: This study demonstrates the feasibility and utility of a novel machine learning-based PEP risk estimation tool with high negative predictive value to aid in prophylaxis selection and identify patients at low risk who may not require extended post-procedure monitoring.

Current and Emerging Applications of Artificial Intelligence (AI) in the Management of Pancreatobiliary (PB) disorders.

PURPOSE OF REVIEW: In this review, we aim to summarize the existing literature and future directions on the use of artificial intelligence (AI) for the diagnosis and treatment of PB (pancreaticobiliary) disorders. RECENT FINDINGS: AI models have been developed to aid in the diagnosis and management of PB disorders such as pancreatic adenocarcinoma (PDAC), pancreatic neuroendocrine tumors (pNETs), acute pancreatitis, chronic pancreatitis, autoimmune pancreatitis, choledocholithiasis, indeterminate biliary strictures, cholangiocarcinoma and endoscopic procedures such as ERCP, EUS, and cholangioscopy. Recent studies have integrated radiological, endoscopic and pathological data to develop models to aid in better detection and prognostication of these disorders. AI is an indispensable proponent in the future practice of medicine. It has been extensively studied and approved for use in the detection of colonic polyps. AI models based on clinical, laboratory, and radiomics have been developed to aid in the diagnosis and management of various PB disorders and its application is ever expanding. Despite promising results, these AI-based models need further external validation to be clinically applicable.

Development of a novel CT-based index for predicting the number of extracorporeal shockwave lithotripsy (ESWL) sessions required for successful fragmentation of obstructing pancreatic duct stones.

BACKGROUND & AIM: Extracorporeal shock wave lithotripsy (ESWL) is used for the treatment of pancreatic duct stones (PDS) in patients with chronic pancreatitis (CP). We aimed to develop a CT based index to predict the required number of ESWL sessions for technical success. METHODS: We retrospectively evaluated patients with PDS secondary to CP who underwent ESWL. Technical success was defined as the complete fragmentation of stones to <3 mm. CT features including PDS size, number, location, and density in Hounsfield units (HU) were noted. We analyzed the relationship between PDS characteristics and the number of ESWL sessions required for technical success. A multiple linear regression model was used to combine size and density into the pancreatic duct stone (PDS) index that was translated into a web-based calculator. RESULTS: There were 206 subjects (mean age 38.6 ± 13.7 years, 59.2% male) who underwent ESWL. PDS size showed a moderate correlation with the number of ESWL sessions (r = 0.42, p < 0.01). PDS in the head required a fewer number of sessions in comparison to those in the body (1.4 ± 0.6 vs. 1.6 ± 0.7, p = 0.01). There was a strong correlation between PDS density and the number of ESWL sessions (r = 0.617, p-value <0.01). The PDS index {0.3793 + [0.0009755 x PDS density (HU)] + [0.02549 x PDS size (mm)]} could accurately predict the required number of ESWL sessions with an AUC of 0.872 (p < 0.01). CONCLUSION: The PDS index is a useful predictor of the number of ESWL sessions needed for technical success that can help in planning and patient counseling.

A machine learning-based choledocholithiasis prediction tool to improve ERCP decision making: a proof-of-concept study.

BACKGROUND: Previous studies demonstrated limited accuracy of existing guidelines for predicting choledocholithiasis, leading to overutilization of endoscopic retrograde cholangiopancreatography (ERCP). More accurate stratification may improve patient selection for ERCP and allow use of lower-risk modalities. METHODS: A machine learning model was developed using patient information from two published cohort studies that evaluated performance of guidelines in predicting choledocholithiasis. Prediction models were developed using the gradient boosting model (GBM) machine learning method. GBM performance was evaluated using 10-fold cross-validation and area under the receiver operating characteristic curve (AUC). Important predictors of choledocholithiasis were identified based on relative importance in the GBM. RESULTS: 1378 patients (mean age 43.3 years; 61.2% female) were included in the GBM and 59.4% had choledocholithiasis. Eight variables were identified as predictors of choledocholithiasis. The GBM had accuracy of 71.5% (SD 2.5%) (AUC 0.79 [SD 0.06]) and performed better than the 2019 American Society for Gastrointestinal Endoscopy (ASGE) guidelines (accuracy 62.4% [SD 2.6%]; AUC 0.63 [SD 0.03]) and European Society of Gastrointestinal Endoscopy (ESGE) guidelines (accuracy 62.8% [SD 2.6%]; AUC 0.67 [SD 0.02]). The GBM correctly categorized 22% of patients directed to unnecessary ERCP by ASGE guidelines, and appropriately recommended as the next management step 48% of ERCPs incorrectly rejected by ESGE guidelines. CONCLUSIONS: A machine learning-based tool was created, providing real-time, personalized, objective probability of choledocholithiasis and ERCP recommendations. This more accurately directed ERCP use than existing ASGE and ESGE guidelines, and has the potential to reduce morbidity associated with ERCP or missed choledocholithiasis.

Improving Pancreatic Cyst Management: Artificial Intelligence-Powered Prediction of Advanced Neoplasms through Endoscopic Ultrasound-Guided Confocal Endomicroscopy.

Despite the increasing rate of detection of incidental pancreatic cystic lesions (PCLs), current standard-of-care methods for their diagnosis and risk stratification remain inadequate. Intraductal papillary mucinous neoplasms (IPMNs) are the most prevalent PCLs. The existing modalities, including endoscopic ultrasound and cyst fluid analysis, only achieve accuracy rates of 65-75% in identifying carcinoma or high-grade dysplasia in IPMNs. Furthermore, surgical resection of PCLs reveals that up to half exhibit only low-grade dysplastic changes or benign neoplasms. To reduce unnecessary and high-risk pancreatic surgeries, more precise diagnostic techniques are necessary. A promising approach involves integrating existing data, such as clinical features, cyst morphology, and data from cyst fluid analysis, with confocal endomicroscopy and radiomics to enhance the prediction of advanced neoplasms in PCLs. Artificial intelligence and machine learning modalities can play a crucial role in achieving this goal. In this review, we explore current and future techniques to leverage these advanced technologies to improve diagnostic accuracy in the context of PCLs.

Pancreatic duct pressure: A review of technical aspects and clinical significance.

Pancreatic duct pressure (PDP) dynamics comprise an intricately modulated system that helps maintain homeostasis of pancreatic function. It is affected by various factors, including the rate of pancreatic fluid secretion, patency of the ductal system, sphincter of Oddi function, and pancreatic fluid characteristics. Disease states such as acute and chronic pancreatitis can alter the normal PDP dynamics. Ductal hypertension or increased PDP is suspected to be involved in the pathogenesis of pancreatic pain, endocrine and exocrine pancreatic insufficiency, and recurrent pancreatitis. This review provides a comprehensive appraisal of the available literature on PDP, including the methods used in the measurement and clinical implications of elevated PDP.

Factors Associated With Serial Lipase Measurement in Hospitalized Patients With Acute Pancreatitis.

OBJECTIVES: To determine the factors associated with serial lipase measurement in patients with acute pancreatitis (AP). METHODS: Patients admitted to Johns Hopkins Health System between September 2019 and August 2020 with lipase ≥3 times upper limit normal were prospectively identified. Acute pancreatitis was defined using revised Atlanta criteria. Serial lipase measurement was defined as >2 lipase measurements on consecutive days within 7 days of presentation. RESULTS: There were 294 patients with AP with mean age 52.4 ± 16 years (SD), and 155 (52.7%) were male. A total of 227 (77.2%) were admitted to a medical service. There were 111 (37.7%) who underwent serial lipase measurements. There were 89 (30.8%), 36 (12.2%), 6 (1%), and 40 (13.6%) patients with systemic inflammatory response syndrome at time of initial lipase measurement, persistent organ failure, necrosis on admission, and intensive care unit admission. Serial lipase measurements were more likely to be obtained in patients admitted to surgical services (odds ratio, 4.3; 95% confidence interval, 1.4-13.2; P = 0.01) and nontertiary hospitals (odds ratio, 1.8; 95% confidence interval, 1.0-2.9; P = 0.04). CONCLUSION: More than one-third of AP patients undergo serial lipase measurements. This practice is more likely to occur on surgical services and in nontertiary hospitals.

Postendoscopic Retrograde Cholangiopancreatography Pancreatitis Pathophysiology and Prevention.

Endoscopic retrograde cholangiopancreatography (ERCP) is an essential procedure for the management of pancreato-biliary disorders. Pancreatitis remains the most frequent complication of the ERCP procedure, and it is, therefore, necessary to recognize the pathophysiology and risk factors contributing to the development of pancreatitis and understand the methods to prevent this complication.

Colonoscopy 3D video dataset with paired depth from 2D-3D registration.

Screening colonoscopy is an important clinical application for several 3D computer vision techniques, including depth estimation, surface reconstruction, and missing region detection. However, the development, evaluation, and comparison of these techniques in real colonoscopy videos remain largely qualitative due to the difficulty of acquiring ground truth data. In this work, we present a Colonoscopy 3D Video Dataset (C3VD) acquired with a high definition clinical colonoscope and high-fidelity colon models for benchmarking computer vision methods in colonoscopy. We introduce a novel multimodal 2D-3D registration technique to register optical video sequences with ground truth rendered views of a known 3D model. The different modalities are registered by transforming optical images to depth maps with a Generative Adversarial Network and aligning edge features with an evolutionary optimizer. This registration method achieves an average translation error of 0.321 millimeters and an average rotation error of 0.159 degrees in simulation experiments where error-free ground truth is available. The method also leverages video information, improving registration accuracy by 55.6% for translation and 60.4% for rotation compared to single frame registration. 22 short video sequences were registered to generate 10,015 total frames with paired ground truth depth, surface normals, optical flow, occlusion, six degree-of-freedom pose, coverage maps, and 3D models. The dataset also includes screening videos acquired by a gastroenterologist with paired ground truth pose and 3D surface models. The dataset and registration source code are available at https://durr.jhu.edu/C3VD.

Frequent guidewire passage into the pancreatic duct is an independent risk factor for postendoscopic retrograde cholangiopancreatography (ERCP) pancreatitis (PEP) among high-risk individuals: A post-hoc analysis of a randomized controlled trial data.

OBJECTIVES: To systematically evaluate the patient and procedural risk factors for post-endoscopic retrograde cholangiopancreatography (ERCP) pancreatitis (PEP) among patients receiving rectal indomethacin. METHODS: Data from a randomized controlled trial (RCT) of high-risk patients undergoing ERCP who received rectal indomethacin with or without topical epinephrine was evaluated. PEP was defined based on the consensus criteria. Pancreatic stenting was excluded to avoid confounding results with the role of epinephrine spray. Multivariable logistic regression analysis was used to identify patient and procedural risk factors for PEP. RESULTS: Among 960 patients enrolled in the RCT, the PEP incidence was 6.4%. An increased risk of PEP was seen with age <50 years and female gender (odds ratio [OR] 2.40, 95% confidence interval [CI] 1.35-4.26), malignant biliary stricture(s) (OR 3.51, 95% CI 1.52-8.10), >2 guidewire passes into the pancreatic duct (PD) (OR 2.84, 95% CI 1.43-5.64), and pancreatic brush cytology (OR 6.37, 95% CI 1.10-36.90), whereas a decreased risk of PEP was seen with contrast- over guidewire-assisted cannulation (OR 0.14, 95% CI 0.02-0.99) and the use of lactated Ringer's (LR) over other fluid types (OR 0.52, 95% CI 0.27-0.98). There was a significant trend between the number of guidewire passes into the PD and PEP risk (P = 0.002). CONCLUSIONS: More than two guidewire passes into the PD and pancreatic brush cytology increased while the use of LR decreased the risk of PEP among high-risk patients receiving rectal indomethacin. Pancreatic stent placement and/or LR should be considered in patients with >2 guidewire passes into the PD.

Lactated Ringers Use in the First 24 Hours of Hospitalization Is Associated With Improved Outcomes in 999 Patients With Acute Pancreatitis.

INTRODUCTION: Recent pilot trials in acute pancreatitis (AP) found that lactated ringers (LR) usage may result in decreased risk of moderately severe/severe AP compared with normal saline, but their small sample sizes limit statistical power. We investigated whether LR usage is associated with improved outcomes in AP in an international multicenter prospective study. METHODS: Patients directly admitted with the diagnosis of AP were prospectively enrolled at 22 international sites between 2015 and 2018. Demographics, fluid administration, and AP severity data were collected in a standardized prospective manner to examine the association between LR and AP severity outcomes. Mixed-effects logistic regression analysis was performed to determine the direction and magnitude of the relationship between the type of fluid administered during the first 24 hours and the development of moderately severe/severe AP. RESULTS: Data from 999 patients were analyzed (mean age 51 years, female 52%, moderately severe/severe AP 24%). Usage of LR during the first 24 hours was associated with reduced odds of moderately severe/severe AP (adjusted odds ratio 0.52; P = 0.014) compared with normal saline after adjusting for region of enrollment, etiology, body mass index, and fluid volume and accounting for the variation across centers. Similar results were observed in sensitivity analyses eliminating the effects of admission organ failure, etiology, and excessive total fluid volume. DISCUSSION: LR administration in the first 24 hours of hospitalization was associated with improved AP severity. A large-scale randomized clinical trial is needed to confirm these findings.

Interpretation of serum pancreatic enzymes in pancreatic and nonpancreatic conditions.

PURPOSE OF REVIEW: Serum levels of amylase and lipase can be elevated in nonpancreatic conditions that may or may not be associated with abdominal pain. This leads to a large proportion of patients being falsely labeled as having acute pancreatitis. In this review, we aim to summarize the existing evidence on pancreatic enzyme elevation in various pancreatic and nonpancreatic conditions and its practical implications in clinical practice and healthcare. RECENT FINDINGS: Serum amylase and lipase levels are not specific for pancreatitis. Attempts have been made to validate newer biomarkers including pancreatic elastase, serum trypsin, urinary trypsinogen-activated peptide, phospholipase A2, carboxypeptidase B, activated peptide of carboxypeptidase B, the trypsin 2 alpha 1 activation complex, and circulating cell-free DNA for the diagnosis of acute pancreatitis. SUMMARY: Serum lipase levels can be elevated in many intra-abdominal inflammatory conditions. Although more sensitive and specific than amylase, serum lipase levels are not sufficient to diagnose acute pancreatitis in patients with abdominal pain. There is a need to increase stress on radiological evidence as well increase cut-off levels of enzyme elevation for a more accurate diagnosis of acute pancreatitis.

Incidence, severity, and mortality of post-ERCP pancreatitis: an updated systematic review and meta-analysis of 145 randomized controlled trials.

BACKGROUND AND AIMS: The incidence, severity, and mortality of post-ERCP pancreatitis (PEP) largely remain unknown with changing trends in ERCP use, indication, and techniques. We sought to determine the incidence, severity, and mortality of PEP in consecutive and high-risk patients based on a systemic review and meta-analysis of patients in placebo and no-stent arms of randomized control trials (RCTs). METHODS: The MEDLINE, Embase, and Cochrane databases were searched from the inception of each database to June 2022 to identify full-text RCTs evaluating PEP prophylaxes. The incidence, severity, and mortality of PEP from the placebo or no-stent arms of RCTs were recorded for consecutive and high-risk patients. A random-effects meta-analysis for a proportions model was used to calculate PEP incidence, severity, and mortality. RESULTS: One hundred forty-five RCTs were found with 19,038 patients in the placebo or no-stent arms. The overall cumulative incidence of PEP was 10.2% (95% confidence interval [CI], 9.3-11.3), predominantly among the academic centers conducting such RCTs. The cumulative incidences of severe PEP and mortality were .5% (95% CI, .3-.7) and .2% (95% CI, .08-.3), respectively, across 91 RCTs with 14,441 patients. The cumulative incidences of PEP and severe PEP were 14.1% (95% CI, 11.5-17.2) and .8% (95% CI, .4-1.6), respectively, with a mortality rate of .2% (95% CI, 0-.3) across 35 RCTs with 3733 patients at high risk of PEP. The overall trend for the incidence of PEP among patients randomized to placebo or no-stent arms of RCTs has remained unchanged from 1977 to 2022 (P = .48). CONCLUSIONS: The overall incidence of PEP is 10.2% but is 14.1% among high-risk patients based on this systematic review of placebo or no-stent arms of 145 RCTs; this rate has not changed between 1977 and 2022. Severe PEP and mortality from PEP are relatively uncommon.

Effective, safe and efficient porcine model of Forrest Ib bleeding gastric and colonic ulcers.

BACKGROUND: Developing effective gastrointestinal (GI) bleeding animal models is necessary to advance endoscopic hemostasis methods and train endoscopists on their use. Our aim, therefore, was to develop an effective and safe porcine GI bleeding model in the stomach and colon of large and small-sized oozing-type ulcers. METHODS: Gastric and colonic bleeding ulcers were created using either a hybrid endoscopic submucosal dissection (ESD) technique or a cap-assisted endoscopic mucosal resection (EMR-C) technique in 14 pigs. Prior to ulcer creation, animals were treated with either oral apixaban or intravenous (IV) unfractionated heparin anticoagulation in combination with clopidogrel and aspirin. The primary outcome was the technical success of inducing oozing-type Forrest Ib bleeding ulcers. Secondary outcomes included ulcer diameter, number, creation time and the number of complications associated with each technique. RESULTS: Using hybrid ESD and IV heparin anticoagulation, bleeding was observed in 21/23 (91.3%) gastric ulcers and 6/7 (85.7%) colonic ulcers created. The mean diameter and ulcer creation time were 2.3 ± 0.3 cm and 5.3 ± 0.5 min, respectively, for gastric ulcers and 2.2 ± 0.4 cm and 4.06 ± 0.6 min, respectively, for colonic ulcers. Using EMR-C and IV heparin anticoagulation, bleeding was observed in 14/15 (93.3%) gastric ulcers and 6/6 (100%) colonic ulcers created. The mean diameter and ulcer creation time were 0.8 ± 0.2 cm and 2.1 ± 0.5 min, respectively, for gastric ulcers and 0.7 ± 0.2 cm and 1.7 ± 0.3 min, respectively, for colonic ulcers. None of the ulcers created in animals anticoagulated with apixaban developed bleeding. None of the 14 pigs developed any complications. CONCLUSION: We have demonstrated the effectiveness and safety of a porcine GI bleeding model utilizing IV heparin anticoagulation and either hybrid ESD or EMR-C techniques to create oozing-type bleeding ulcers in the stomach and colon with customizable size.

Cystic fibrosis transmembrane conductance regulator modulators and the exocrine pancreas: A scoping review.

BACKGROUND: Cystic fibrosis transmembrane conductance regulator (CFTR) modulators improve pulmonary outcomes in subjects with cystic fibrosis (CF); however, the effects on pancreatic manifestations are not well characterized. We hypothesized that CFTR modulators would improve measures of exocrine pancreatic function and outcomes. METHODS: We performed a systematic search to identify studies reporting measures of the exocrine pancreas in humans treated with CFTR modulators. Only studies reporting baseline and on-treatment assessments were included. RESULTS: Of 630 identified studies, 41 met inclusion criteria. CFTR modulators reduced acute pancreatitis events by 85% overall (rate ratio 0.15, 95% confidence interval (CI) 0.04, 0.52), with a greater effect seen in the subgroup with pancreas sufficient CF (PS-CF) (rate ratio 0.13 (95% CI 0.03, 0.53). Among 293 subjects with baseline and on-treatment evaluation of pancreas sufficiency, 253 were pancreas insufficient at baseline and 54 (21.3%) converted to pancreas sufficiency. Of 32 subjects with baseline FE-1 values <200 mcg/g, 16 (50%) increased to ≥200 mcg/g. Serum trypsin decreased by a mean of 565.9 ng/mL (standard deviation (SD) 311.8), amylase decreased by 38.2 U/L (SD 57.6), and lipase decreased by 232.3 U/L (SD 247.7). CONCLUSIONS: CFTR modulator use reduces acute pancreatitis frequency and improves indirect measures of exocrine pancreas function. Future interventional studies that evaluate the mechanism and impact of CFTR modulators on acute pancreatitis and pancreas sufficiency in patients with CFTR dysfunction are warranted.

Rationale and Design for the Diabetes RElated to Acute Pancreatitis and Its Mechanisms Study: A Prospective Cohort Study From the Type 1 Diabetes in Acute Pancreatitis Consortium.

ABSTRACT: Acute pancreatitis (AP) is a disease characterized by an acute inflammatory phase followed by a convalescent phase. Diabetes mellitus (DM) was historically felt to be a transient phenomenon related to acute inflammation; however, it is increasingly recognized as an important late and chronic complication. There are several challenges that have prevented precisely determining the incidence rate of DM after AP and understanding the underlying mechanisms. The DREAM (Diabetes RElated to Acute Pancreatitis and its Mechanisms) Study is a prospective cohort study designed to address these and other knowledge gaps to provide the evidence needed to screen for, prevent, and treat DM after AP. In the following article, we summarize literature regarding the epidemiology of DM after AP and provide the rationale and an overview of the DREAM study.