Journal of clinical gastroenterology最新文献

Background: Early recurrence of intrahepatic cholangiocarcinoma (ICC) is difficult to predict. Traditional machine learning prediction models, characterized by their black-box nature, may be biases or ethical risks.

Methods: The XGBoost algorithm develops the machine learning prediction model. The area under the receiver operating characteristic curve (AUC) served for evaluating model performance. The SHAP algorithm conducts interpretability analysis.

Results: A total of 503 patients with 323 in the training cohort and 180 in the validation cohort. Tumor size, lymph node metastasis, microvascular invasion (MVI), and CA19-9 levels were identified as independent predictors of ICC early recurrence. The predictive model demonstrated the highest discriminative power in both training and validation cohorts (AUC 0.76 vs. 0.72, respectively). SHAP analysis demonstrates the decision-making process of the machine learning model.

Conclusions: The XGBoost model for predicting early recurrence of ICC demonstrates accuracy and reliability. Explainable machine learning models, which balance transparency and accuracy.

Goals: To compare the effectiveness, durability, and safety of vedolizumab and upadacitinib for CD through 52 weeks.

Background: Comparative real-world data for vedolizumab versus upadacitinib in Crohn's disease (CD) are limited.

Study: This retrospective cohort study included 139 adults with active CD who began vedolizumab (n=72) or upadacitinib (n=67) during 2023 at a large academic health system. Co-primary outcomes were steroid-free clinical remission (SFCR) at 12 and 52 weeks and treatment discontinuation within 52 weeks; secondary outcomes included clinical response at 12 and 52 weeks. Inverse probability of treatment weighting balanced relevant confounders. Logistic regression was used for binary outcomes and Cox proportional hazards and competing risks regression were used for treatment discontinuation. Adverse events were ascertained by manual chart review.

Results: After weighting, all covariates were balanced (standardized mean differences <0.10). At 12 weeks, vedolizumab was associated with lower odds of clinical response versus upadacitinib (OR: 0.36; 95% CI: 0.16-0.85). There were no significant differences for SFCR, treatment discontinuation, or other outcomes through 52 weeks. Competing risks regression, accounting for adverse events as competing events, showed a higher incidence of treatment discontinuation due to nonresponse for vedolizumab, but this did not reach statistical significance. Adverse events within 52 weeks were comparable (vedolizumab 33% vs. upadacitinib 39%; P=0.45), and discontinuations due to adverse events were infrequent (3% vs. 6%).

Conclusions: In this tertiary-center cohort, upadacitinib produced faster clinical response at 12 weeks, but SFCR, durability, and safety profiles were similar through 52 weeks.

Objective: To identify risk factors for short-term mortality in severe acute pancreatitis (SAP), establish a predictive model for early high-risk patient identification, and guide clinical decision-making.

Methods: SAP patients admitted to the Affiliated Hospital of Xuzhou Medical University from September 2018 to September 2025 were enrolled, divided into mortality and survival groups by 28-day prognosis. Clinical data were collected. Features were strictly selected through Least Absolute Shrinkage and Selection Operator (LASSO) regression, Boruta algorithm, and Recursive Feature Elimination (RFE). Seven machine learning (ML) models were built, with external validation using Medical Information Mart for Intensive Care IV (MIMIC-IV) data. Model performance was evaluated through receiver operating characteristic (ROC) curves, calibration curves, and decision curves. SHapley Additive exPlanations (SHAP) analysis was used to interpret contributions of important features, and a web-based calculator was developed for visualization.

Results: Ten features were selected. The Gradient Boosting Machine (GBM) model had the best generalization, with area under the ROC curve (AUC) values of 0.964 (95% CI: 0.942-0.987, training), 0.927 (95% CI: 0.885-0.970, testing), and 0.811 (95% CI: 0.772-0.851, validation). Calibration curves confirmed predicted-actual consistency; decision curves showed net clinical benefit. SHAP analysis identified the ranking of feature importance as follows: mechanical ventilation, age, blood urea nitrogen, urine output, lactate, total bilirubin, platelet count, congestive heart failure history, red blood cell distribution width, and serum creatinine. The web-based calculator had good clinical applicability.

Conclusion: The GBM model demonstrates the best performance in predicting short-term mortality in SAP patients.

Background and aims: Psychiatric disorders are more common in inflammatory bowel disease (IBD) patients compared with the general population. We aimed to investigate the prevalence and risk factors of depression, schizophrenia, and suicide attempts among IBD patients in Israel.

Methods: Data on IBD patients over 25 years (1999 to 2024) was extracted from the Clalit Health Maintenance Organization. Possible risk factors were investigated via multivariate analyses. A matched case-control analysis was performed of attempted suicide cases.

Results: A total of 3,262,623 adults were analyzed. The frequency rate of suicide attempts was 1.5%, schizophrenia 1.3%, and depression 25.7% among 19,100 Crohn's disease (CD) patients compared with 1.1%, 1.4%, and 23.0% among 17,731 ulcerative colitis (UC) patients. In a multivariate analysis, female gender (OR 1.284, P <0.001), smoking (OR 1.479, P <0.001), CD (OR 1.274, P <0.010), schizophrenia (OR 7.69, P <0.001), and depression (OR 6.284, P <0.001) were found to be predictors of suicide attempts. Case-control analysis of 490 patients with a prior suicide attempt revealed significantly higher rates of schizophrenia (13.1% vs. 1.2%, P <0.001) and depression (70.0% vs. 19.6%, P <0.001) among the suicide group. In a multivariate analysis, schizophrenia (OR 9.873, P <0.001) and depression (OR 8.964, P <0.001) were significant risk factors for suicide attempts. The mortality rate among those with a suicide attempt (98/490, 20%) was significantly higher compared with those without a suicide attempt (5301/36,341, 14.6%, P <0.001).

Conclusion: The main risk factor for suicide attempts among IBD patients is comorbid psychiatric disease. Proactive mental health care by gastroenterologists and integrated psychology services is highly recommended.

Objective: To further enhance the evaluation capabilities of Macroscopic On-site Evaluation (MOSE) for EUS-FNA, we developed a device featuring an adjustable light source, in the hope of more efficient observation of FNA specimens. Then we modified MOSE with this device and compared it with the traditional MOSE in assessing specimen adequacy and diagnostic accuracy.

Methods: This is a single-center cross-randomized controlled trial. Patients with pancreatic solid masses undergoing EUS-FNA (22G/25G) were enrolled, and 2 punctures were conducted with a different sequence of traditional or modified MOSE. Specimens were classified as insufficient if the macroscopic visible core (MVC) <4 mm. Each puncture was independently submitted for pathology analysis.

Results: A total of 90 patients were included in the study, yielding 180 FNA specimens. Modified MOSE demonstrated greater consistency with pathologic assessments of specimen adequacy compared with traditional MOSE, with accuracy rates of 71.1% and 50%, respectively (P=0.004). The diagnostic accuracy rate of the samples evaluated as adequate by traditional MOSE was 81.0% and 96.8% by modified MOSE (P=0.019).

Conclusion: The use of modified MOSE for 22G/25G FNA specimens allows for a more precise determination of specimen adequacy, thus enhancing diagnostic accuracy and assisting endosonographers in determining the optimal endpoint for puncture.

Background/aims: While commonly prescribed in patients with inadequate bowel preparation, split-dose, 2-day bowel preparation (2DBP) with 8 L of polyethylene glycol-electrolyte lavage solution (PEG-ELS) has unknown effectiveness and safety.

Study aims: To assess the effectiveness of a split-dose, 8 L-2DBP in patients with a prior failed bowel preparation.

Methods: We retrospectively identified all outpatients with inadequate bowel preparation on index colonoscopy with a 1-day split-dose prep who had repeat colonoscopy with a split-dose 2DBP (2015 to 2021) at a single Veterans Affairs medical center. We then assessed adequate bowel preparation (outcome), defined as excellent/good or Boston Bowel Prep Score (BBPS) 6 or greater, with a score of 2 in all colon segments. We also collected data on patient factors associated with inadequate preparation. Logistic regression was used to identify associations between these factors and inadequate preparation after split-dose 2DBP. All patient records were manually reviewed to identify potential adverse effects resulting in emergency room presentation.

Results: Among 400 patients given 2DBP, 83.5% (95% CI: 79.5%-87.0%) achieved an adequate preparation. In multivariable analysis, a higher CCI (OR: 1.14, 95% CI: 1.04-1.25), opioid use (OR: 1.97, 95% CI: 1.03-3.77), and rural home address (OR: 1.87, 95% CI: 1.01-3.46) were associated with inadequate bowel preparation. No attributable adverse events were recorded.

Conclusion: In this difficult-to-prep population, split-dose 2DBP is effective in more than 80% of patients without evidence of severe adverse events. This intensive prep was less effective in patients with greater comorbidity burden, opioid use, or a rural home address.

Metabolic dysfunction-associated steatotic liver disease (MASLD) is a growing global health concern, closely linked to obesity and metabolic syndrome, and characterized by hepatic steatosis, inflammation, and fibrosis. While lifestyle modifications and bariatric surgery for weight loss have been the cornerstones of treatment, endoscopic bariatric therapies (EBTs) have emerged as minimally invasive alternatives that offer meaningful weight loss and metabolic improvements, potentially benefiting patients with MASLD. Current evidence indicates that EBTs can significantly reduce liver steatosis, improve liver biochemistry markers, and enhance overall metabolic health. This review aims to evaluate the current evidence on the efficacy, safety, and potential role of EBTs in the management of MASLD, highlighting their impact on liver-related outcomes and metabolic parameters. A comprehensive literature review was conducted using PubMed, identifying observational studies, systematic reviews, meta-analyses, and retrospective and prospective trials that examine the effects of EBTs on metabolic markers and liver disease. This review underscores the potential of EBTs as therapeutic options for MASLD, particularly for patients who are unable to achieve sufficient weight loss through lifestyle interventions alone. While current evidence supports their potential role, further large-scale, long-term studies are needed to establish their efficacy in altering long-term disease progression and to define their place in clinical management algorithms.

Background: Acute pancreatitis (AP), a common acute abdominal disease, has a high mortality rate in severe cases. Accurate mortality prediction is crucial for clinical decision-making. Machine learning (ML) models have shown potential in predicting AP mortality, aiding clinicians in understanding prediction mechanisms and formulating personalized treatment plans.

Objective: This study evaluates and compares the performance of ML models in predicting early mortality in AP patients to provide evidence for mortality prediction and guide the development of clinical prediction tools.

Methods: A comprehensive search of PubMed, Web of Science, Cochrane Library, and Embase databases was conducted for literature published between January 1, 2012, and April 25, 2025. Effect sizes were synthesized using a random-effects model, and subgroup analyses were performed based on model characteristics to explore result heterogeneity.

Results: Twenty-four studies were included. ML models demonstrated high predictive performance for AP mortality risk. In training sets, the pooled C-index was 0.84 (95% CI: 0.81-0.86), with 0.841 (95% CI: 0.806-0.877) for in-hospital mortality. External validation sets showed a pooled C-index of 0.84 (95% CI: 0.82-0.86) and in-hospital mortality prediction of 0.826 (95% CI: 0.798-0.855). ML models outperformed traditional scoring tools (pooled C-index: 0.754, 95% CI: 0.734-0.775 for standard systems). Common predictors included age, blood urea nitrogen, total bilirubin, white blood cells, hemoglobin, blood pressure, and respiratory rate.

Conclusions: Machine learning demonstrates excellent accuracy in predicting the mortality of AP. This offers a reference for updating or creating a simple clinical prediction tool.

Goal: The aim of our study is to evaluate for contributing factors related to intolerance of wireless pH capsule endoscopy.

Background: Wireless capsule endoscopy has been a key diagnostic modality in the quantification of esophageal acid exposure time. Common complications are wireless capsule retention and postdeployment pain; however, some patients experience significant intolerance after the procedure.

Study: Retrospective study of patients in the adult gastroenterology department who have had a wireless pH monitoring from January 2017 to June 2023. Descriptive statistics and pairwise statistical tests of association were computed for each clinical covariate of interest. In addition, a multivariate logistic regression model was used to jointly infer the relationship between postprocedure pain and numerous covariates.

Results: Of the 204 participants who underwent the wireless pH capsule placement, 18 (8.73%) patients had documented severe postprocedural pain. The covariates retention (P<0.001) and irritable bowel syndrome (P=0.015) were inferred to have statistically significant associations with postprocedure pain. The odds of a patient with a history of depression experiencing postprocedure pain are estimated to be 8.18 (95% CI: 1.53-53.5; P=0.019) times the odds of a patient without depression. The odds ratio of antidepressant use was estimated to be 0.17 (95% CI: 0.03-0.77; P=0.029) times (ie, 83%) less than the odds without antidepressant use.

Conclusions: IBS diagnosis, depression, and capsule esophageal retention were each associated with postprocedural intolerance. Interestingly, antidepressant use may be protective, possibly related to the neuromodulating properties of these medications. Further evaluation is needed to validate this observation.

Introduction: Endoscopic mucosal resection (EMR) is the standard approach for managing large nonpedunculated colorectal polyps (LNPCPs ≥15 mm). Hot EMR (H-EMR) offers low recurrence rates but carries a higher risk of delayed postpolypectomy bleeding (DPPB), while cold EMR (C-EMR) has a more favorable safety profile, but its higher recurrence rates remain a concern. Given these trade-offs, we conducted a cost-effectiveness analysis comparing C-EMR and H-EMR for LNPCPs.

Methods: We conducted an incremental cost-effectiveness analysis over a 6-month time horizon using a decision tree model informed by the pooled data of randomized studies evaluating C-EMR versus H-EMR. Costs-including for EMR, delayed bleeding, and hospitalization-were derived from CMS reimbursement data and published sources. The Incremental Cost-Effectiveness Ratio (ICER) was determined for the base patient undergoing H-EMR versus C-EMR for LNPCPs. Analysis was performed using TreeAge Pro Health care 2024.

Results: Pooled data from RCTs comprising 1516 LNPCPs (766 in C-EMR and 750 in H-EMR group) in 1442 patients were utilized. In the base case of a 66.8-year-old patient undergoing endoscopic resection for LNPCPs, C-EMR was associated with an incremental cost of -$286.67, incremental effectiveness of 0.0004282, resulting in an incremental cost-effectiveness ratio (ICER) of -$669,448 per QALY. This indicates that C-EMR is cost-effective compared with H-EMR at a WTP threshold of $100,000 per QALY.

Conclusion: Our analysis shows that C-EMR is a cost-effective strategy compared with H-EMR for LNPCPs ≥15 mm. While H-EMR offers lower recurrence rates, its higher rates of adverse events-such as DPPB and perforation-contribute to increased costs and reduced overall effectiveness.