Clinical and Translational Gastroenterology最新文献

Introduction: Colorectal cancer (CRC) remains a leading cause of cancer associated death in the United States and colonoscopy the primary screening strategy for prevention. Rates of adenomatous and serrated neoplasia detection are inversely associated with post-colonoscopy CRC. This crucial quality metric depends on accurate ascertainment of colorectal neoplasia findings from both endoscopy and histopathology records. We aimed to assess the feasibility of a random forest machine learning model to rapidly and accurately categorize colorectal neoplasia from electronic health record data.

Methods: A retrospective cohort study compared neoplasia detection rates among individuals undergoing colonoscopy at a large academic institution to develop a rule-based algorithm to categorize colorectal neoplasia from endoscopy reports and pathology SNOMED II codes. This cohort provided a large training set to develop a natural language processing (NLP) system using a random forest approach to automatically classify unstructured pathology findings into adenoma, serrated, or advanced neoplasms. This system was manually validated through an independent holdout set.

Results: The training set comprised 35,953 unstructured pathology reports with matched SNOWMED II codes from 95,188 unstructured colonoscopy reports. The final model was assessed on an independent holdout set of 337 manually annotated procedures obtaining an AUC of 0.997 (CI 0.994 - 1), 0.99 (CI 0.98-1), and 0.99 (CI 0.98-0.99) for prediction of adenoma, serrated, and advanced lesions respectively.

Discussion: The random forest-based hybrid NLP system for classification of colonoscopy results was both accurate and explainable. NLP combined with effective machine learning algorithms can provide a scalable strategy for colonoscopy quality monitoring.

Introduction: Mucosal exposure devices (MEDs) and computer-aided detection (CADe) systems may both improve adenoma detection through distinct mechanisms: expanding mucosal visualization and highlighting lesions, respectively. This study investigated the efficacy of combining CADe-assisted colonoscopy with a MED compared to CADe-assisted colonoscopy alone.

Methods: This international, multicenter, prospective, non-randomized, single-arm study (NTC05220345) was conducted at three centers that also participated in the previous DISCOVERY II randomized controlled trial, comparing CADe-assisted with conventional colonoscopy. Patients referred for diagnostic, non-fecal immunochemical test screening, or surveillance colonoscopy and underwent CADe-assisted colonoscopy (DISCOVERY™, PENTAX Medical) with a MED using an integrated inflatable balloon (G-EYE™, PENTAX Medical). The primary outcome was adenoma detection rate (ADR); secondary outcomes included sessile serrated lesion detection rate (SSLDR) and withdrawal time without interventions. Outcomes were compared to historical controls of the CADe-arm of the DISCOVERY II study.

Results: Of 196 enrolled participants, 182 were included in the final analysis and compared with 250 participants from the historical CADe-arm. ADR was 47.3% in the CADe + MED-group vs. 38.4% in the CADe-group (p=.066; absolute difference: 8.9%, 95% CI: -0.6-18.3). Mixed-effects logistic regression model adjusting for clustering and confounders calculated an odds ratio of 1.16 (95% CI: 0.74-1.81). Median withdrawal time was slightly longer with CADe + MED compared to CADe-only (10.0 vs. 9.2 minutes, p=.004), while SSLDR was not significantly different (12.6% vs. 18.4%, p=.11).

Conclusions: In this study using historical controls, CADe-assisted colonoscopy combined with a MED did not significantly increase ADR compared to CADe alone, suggesting limited synergistic benefit.

Introduction: Entecavir (ETV), tenofovir disoproxil fumarate (TDF), and tenofovir alafenamide (TAF) are first-line nucleos(t)ide analogs (NA) for chronic hepatitis B (CHB). Real-world monitoring of patients on these agents and their comparative renal safety remain poorly characterized. We evaluated guideline-adherent monitoring practices and compared renal dysfunction risk across ETV, TDF, and TAF.

Methods: We retrospectively analyzed patients with CHB who initiated ETV, TDF, or TAF between 2012 and 2022. Levels of serum alanine aminotransferase, total bilirubin, albumin, serum creatinine, hepatitis B virus DNA, and abdominal sonograms were assessed every 6 months during the 36 months of NA treatment. Incidence rates and adjusted hazard ratios (HRs) for renal dysfunction were estimated by Cox regression.

Results: Of the 2,155 enrolled patients, 65.8% received ETV, 23.1% received TDF, and 11.1% underwent TAF. Alanine aminotransferase was monitored in >90% across all groups; other tests (bilirubin, albumin, creatinine, hepatitis B virus DNA, sonogram) were performed in only 20%-80%. After multivariable adjustment, TDF (HR 1.41; 95% confidence interval 0.95-2.08) and TAF (HR 0.91; 95% confidence interval 0.52-2.18) showed no significant difference in renal dysfunction risk vs ETV. Independent predictors of increased renal risk included older age, higher Charlson comorbidity index, fibrosis-4 score, and diuretic use, whereas elevated serum albumin levels were associated with a lower risk.

Discussion: In this real-world cohort, adherence to recommended monitoring for patients with CHB on NAs was suboptimal. ETV, TDF, and TAF demonstrated comparable renal safety profiles over 3 years.

Introduction: Women with endometriosis have a higher risk of developing inflammatory bowel diseases (IBDs). This study aimed to better understanding the impact of endometriosis on the course of IBD.

Methods: We conducted a retrospective cohort study in 18 French and Belgian IBD centers between June 2022 and March 2023. Any patient with both conditions was eligible for inclusion. They were randomly matched to 1 or 2 patients with IBD without endometriosis. The impact on IBD progression was assessed using a composite severity criterion including intestinal damage or need for bowel surgery.

Results: Overall, 207 patients with both conditions (149 Crohn's disease [CD]; 58 ulcerative colitis [UC]) were matched to 409 patients with IBD alone. The median follow-up duration for IBD was 10 years (5.75-17). No difference was observed between the 2 groups regarding CD location, disease phenotype, and anoperineal involvement. Proctitis were more frequent in patients with UC and endometriosis. Patients with IBD with endometriosis were significantly less exposed to immunosuppressants (UC P < 0.01; CD P < 0.001) and biologics (UC P < 0.01; CD P < 0.001). Patients with CD with endometriosis had a less severe disease course compared with patients without endometriosis (hazard ratio 0.68, 95% confidence interval 0.50-0.92, P = 0.011). Patients with UC with endometriosis had not a significant different disease course compared with patients without endometriosis (hazard ratio 1.73, 95% confidence interval 0.74-4.00, P = 0.20). These results were similar in the subgroup of patients with endometriosis treated surgically.

Discussion: Endometriosis does not negatively influence the course of IBD, patients with CD even have a less severe progression. Patients were significantly less exposed to immunosuppressants and biologics.

Introduction: Interleukin-13 receptor alpha 2 (IL13RA2), triggering receptor expressed on myeloid cells-1 (TREM-1), and oncostatin M (OSM) may be associated with response to tumor necrosis factor-α antagonists (TNFAs) in inflammatory bowel disease. We aimed to assess the direction of association between TNFA-induced clinical remission and IL13RA2 and TREM-1, respectively, and assess the value of combining biomarkers for identifying nonresponders.

Methods: Plasma samples from a retrospective inflammatory bowel disease cohort were collected before TNFA start. Clinical remission at 1-year, surgery, hospitalization, adverse drug events, and TNFA discontinuation were assessed. IL13RA2 and TREM-1 concentrations were compared between those with and without 1-year clinical remission. OSM data were obtained from our previous cohort. Where significant, TREM-1 and IL23RA2 thresholds associated with clinical remission at 1-year were assessed using a receiver operating characteristic analysis. Significant biomarkers were combined using a linear discriminant analysis. The performance characteristics were assessed for individual biomarkers and biomarker combinations.

Results: In Crohn's disease (CD) (n = 95) and ulcerative colitis (UC) (n = 53), higher IL13RA2 concentrations, but not TREM-1, were found among those not achieving TNFA-associated clinical remission at 1-year (IL13RA2, CD, P < 0.0001; UC, P = 0.0003). IL13RA2 thresholds, 4.554 ng/mL (CD) and 6.117 ng/mL (UC) separated those with and without clinical remission at 1-year (CD, area under the receiver-operating characteristic curve = 0.80, 95% CI = 0.71-0.90, P < 0.0001; UC, area under the receiver-operating characteristic curve = 0.79, 95% CI = 0.66-0.91, P = 0.0005). In CD, combining IL13RA2 and OSM concentrations enhanced prediction accuracy compared with either biomarker alone and increased the identification of other important clinical outcomes.

Discussion: IL13RA2, but not TREM-1, was associated with TNFA response. In CD, its prediction accuracy improves when combined with OSM.

Introduction: Esophageal motility disorders (EMDs) are common in clinical practice, with a high symptomatic burden and significant impact on the patients' quality of life. High-resolution esophageal manometry (HREM) is the gold standard for the evaluation of functional esophageal disorders. The Chicago Classification offers a standardized approach to HREM. However, HREM remains a complex procedure, both in data analysis and in accessibility. This study aimed to develop and validate machine learning (ML) models to detect EMDs according to the Chicago Classification.

Methods: We retrospectively analyzed 618 HREM examinations from 3 centers (Spain and the United States) using 2 recording systems. Labels were assigned by expert consensus as either disorder present or absent for 2 categories: esophagogastric junction outflow disorders and peristalsis disorders. Several ML models were trained and evaluated. ML classifiers were developed using an 80/20 patient-level stratified split for training/validation and testing. Model selection was guided by internal evaluation through repeated 10-fold cross-validation. Model performance was assessed by accuracy and area under the receiver-operating characteristic curve (AUC-ROC).

Results: The GradientBoostingClassifier model outperformed the remaining ML models with an accuracy of 0.942 ± 0.015 and an AUC-ROC of 0.921 ± 0.041 for identifying disorders of esophagogastric junction outflow. The xGBClassifier model detected disorders of peristalsis with an accuracy of 0.809 ± 0.029 and an AUC-ROC of 0.871 ± 0.027. Performance was consistent across repeated validations, demonstrating model robustness and generalization.

Discussion: This multicenter, multidevice study demonstrates that ML models can accurately detect EMDs in HREM. Artificial intelligence-driven HREM may improve diagnosis by standardizing interpretation and reducing interobserver variability.

Introduction: Esophagogastric varices (EGV) are known to correlate with a poorer prognosis in patients with liver cirrhosis or hepatocellular carcinoma. However, their clinical significance in patients with cholangiocarcinoma (CCA) remains unknown. The aim of this study was to investigate the impact of EGV on the outcomes of patients with CCA.

Methods: This retrospective study enrolled 923 consecutive treatment-naive patients diagnosed with CCA between January 2013 and December 2023. Among these, 321 patients received esophagogastroduodenoscopy at the time of CCA diagnosis. The primary end point was to assess the impact of EGV on overall survival (OS) in patients with CCA, whereas the secondary end point was to identify the predictive factors for the occurrence of EGV.

Results: Of the patients analyzed, 47 (14.6%) were diagnosed with EGV by esophagogastroduodenoscopy. Among these, 39 patients did not receive primary prophylaxis for EGV bleeding and were classified as the EGV group, whereas the remaining 274 patients (85.4%) formed the non-EGV group. The median OS was shorter in the EGV group than that in the non-EGV group (182 vs 357 days, P = 0.009). Multivariate analyses identified the presence of EGV as an independent risk factor of poorer OS (hazard ratio 1.823, confidence interval 1.248-2.663, P = 0.002). Besides, fibrosis-4 scores >2.67 and albumin-bilirubin grades >1 were predictive factors for EGV occurrence.

Discussion: While the prevalence of concurrent EGV in patients with CCA was relatively low, its presence was associated with a poorer prognosis. The fibrosis-4 scores and albumin-bilirubin grades predicted the occurrence of EGV.

Introduction: Up to one-third of the patients diagnosed with common variable immunodeficiency (CVID) may develop gastrointestinal (GI) and hepatic manifestations. This study aimed to evaluate the prognostic significance of enteropathy and liver disease in patients with CVID.

Methods: We conducted a retrospective study including all consecutive adult patients with CVID followed in a tertiary care center in Spain from January 1990 to January 2023. A diagnosis of CVID-associated enteropathy (CVID-E) and CVID-associated liver involvement (CVID-L) was established when objective clinical, endoscopic, histologic, radiologic or hemodynamic findings were present. Relevant prognostic outcomes and their risk factors were studied, including survival, GI infections, and GI cancer.

Results: Eighty-nine patients with confirmed CVID were included, 26 of them (29.2%) had CVID-E and 23 (25.8%) had CVID-L. Nineteen (73.1%) patients with CVID-E suffered from GI infections, while 12 (46.2%) presented concurrent liver involvement. In comparison with the rest of the cohort, patients with CVID-E had more frequently liver involvement, GI infections, and GI cancer. Multivariate analysis identified CVID-E as an independent risk factor for GI infections. Twelve (52.2%) patients with CVID-L concurrently exhibited CVID-E, and patients with CVID-L presented more CVID-E, splenomegaly, and a trend toward more GI cancer and GI infections. CVID-L and age at CVID diagnosis emerged as independent risk factors for mortality.

Discussion: GI and hepatic involvements are common in patients with CVID and frequently occur together. These manifestations significantly affect the disease course, increasing the risk of GI infections, GI malignancy, and, in the case of liver disease, mortality.

Introduction: Esophageal adenocarcinoma (EAC) is an aggressive cancer with poor prognosis. Barrett's esophagus (BE) is a critical precursor of EAC. Patients with BE undergo endoscopic surveillance to monitor disease progression although only a small fraction develop EAC. These procedures are invasive and have limited accuracy in predicting BE progression. We evaluated the utility of an 8-protein mass spectrometry panel in predicting progression in patients with BE.

Methods: Eighty untreated controls and 20 cases were selected from our institutional tissue registry. Quantitative mass-spectrometry was performed on microdissected tissue sections. Data were split into 80% training and 20% test sets. We used Least Absolute Shrinkage and Selection Operator-regularized regression to train a logistic classifier on training data. Classifier performance was evaluated in test data.

Results: Ninety-two samples had sufficient tissue for mass spectrometry analysis (18 progressors, 74 nonprogressors). The multivariable regression model produced a sensitivity of 100% and a specificity of 39% in the overall cohort, with AUCs of 0.75 and 0.89 in the overall and test cohorts, respectively. Cox proportional hazards time-to-progression (TTP) showed a hazard ratio of 66.1 (95% CI 7.79-561, P = 0.00012) for the model prediction.

Discussion: The promising performance of the model generated here suggests that the test may aid in selecting patients most likely to benefit from active BE surveillance. Moreover, the association of this model's prediction with time-to-progression may offer decision support for management of patients likely to progress quickly. These results support continued development of this proteomic panel as a risk stratification tool for patients with BE.