BMC Gastroenterology最新文献

Background: Although capsule endoscopy (CE) is a crucial tool for diagnosing small bowel diseases, the need to process a vast number of images imposes a significant workload on physicians, leading to a high risk of missed diagnoses. This study aims to develop an artificial intelligence (AI) model and application based on convolutional neural networks that can automatically recognize various lesions in small bowel capsule endoscopy.

Methods: Three small bowel capsule endoscopy datasets were used for AI model training, validation, and testing, encompassing 12 categories of images. The model's performance was evaluated using metrics such as AUC, sensitivity, specificity, precision, accuracy, and F1 score to select the best model. A human-machine comparison experiment was conducted using the best model and endoscopists with varying levels of experience. Model interpretability was analyzed using Grad-CAM and SHAP techniques. Finally, a clinical application was developed based on the best model using PyQt5 technology.

Results: A total of 34,303 images were included in this study. The best model, MobileNetv3-large, achieved a weighted average sensitivity of 87.17%, specificity of 98.77%, and an AUC of 0.9897 across all categories. The application developed based on this model performed exceptionally well in comparison with endoscopists, achieving an accuracy of 87.17% and a processing speed of 75.04 frames per second, surpassing endoscopists of varying experience levels.

Conclusion: The AI model and application developed based on convolutional neural networks can quickly and accurately identify 12 types of small bowel lesions. With its high sensitivity, this system can effectively assist physicians in interpreting small bowel capsule endoscopy images.Future studies will validate the AI system for video evaluations and real-world clinical integration.

Background: The Rome IV criteria for functional dyspepsia (FD) has strict requirements for symptom frequency and onset duration, making it difficult for patients to meet these criteria in clinical practice. This study aimed to investigate the impact of relaxing the Rome IV criteria on the diagnosis and symptom pattern of FD.

Methods: A cross-sectional, multi-center study was conducted involving 2935 consecutive broadly defined FD patients without positive findings on upper gastrointestinal endoscopy or routine examinations. Questionnaires were used to collect demographic and upper gastrointestinal symptom data. Symptom pattern was compared between Rome IV criteria defined FD patients and those defined by relaxed Rome IV criteria.

Results: Only 22.2% of broadly defined FD patients rigorously fulfilled Rome IV criteria. No significant difference was found for proportion of patients with dyspeptic symptoms, dysmotility-like symptoms, reflux-like symptoms, as well as severity and onset frequency of dyspeptic symptoms (all P > 0.05), between patients who didn't fulfill Rome IV criteria for FD solely due to a duration of 3-6 months and Rome IV criteria defined FD patients. Patients with broadly defined postprandial distress syndrome (PDS) who didn't fulfill Rome IV criteria solely due to a symptom frequency of 1-2 days per week had significantly lower symptom severity (P < 0.001), but similar postprandial symptom characteristics compared to those defined by the Rome IV criteria.

Conclusions: A symptom duration criterion of 3 months may be sufficient for diagnosing FD. Reducing the symptom onset frequency to no less than 1 day per week in the Rome IV criteria for PDS does not affect its postprandial symptom characteristics.

Background: This study explored the correlation between peripheral blood lipid levels and clinicopathological parameters in patients with advanced gastric cancer (GC), focusing on changes in lipid levels during disease progression.

Methods: Pathological features and serum lipid profiles of 179 patients with stage III-IV gastric adenocarcinoma were analyzed. Lipid parameters examined included total cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), very low-density lipoprotein cholesterol (VLDL-C), apolipoprotein AI (Apo AI), apolipoprotein B (Apo B), lipoprotein(a) (Lp(a)), among others. The total cholesterol-lymphocyte score (TL score) and BMI were also calculated. The association between lipid parameters and clinicopathological characteristics such as age, gender, family history, and metastasis sites was assessed.

Results: In GC patients, females had higher TG levels than males. Patients with peritoneal metastasis had significantly lower levels of TC, LDL-C, Apo B, and B/A ratio. Those with lung metastasis exhibited higher LDL-C levels and lower levels of VLDL-C. No significant associations were found between lipid levels and metastasis to distant lymph nodes, liver, or bone. Female patients with ovarian metastasis had significantly lower VLDL-C levels. Multivariate analysis revealed low TC as an independent risk factor for peritoneal metastasis, high LDL-C and low VLDL-C levels for lung metastasis, and younger age and low VLDL-C for ovarian metastasis.

Conclusion: Specific blood lipid levels are significantly associated with metastatic sites in advanced gastric cancer. Lipid profiles could serve as potential biomarkers for predicting metastatic sites in GC patients.

Background: The triglyceride to high density lipoprotein cholesterol ratio (TG/HDL-C) is a confirmed predictive factor for insulin resistance and is suggested to be closely related to metabolic dysfunction-associated fatty liver disease (MAFLD), but previous research is inconclusive. The association between TG/HDL-C and MAFLD incidence was further explored in this large-sample, long-term retrospective cohort study.

Methods: Individuals who participated in the Kailuan Group health examination from July 2006 to December 2007 (n = 49,518) were included. Data from anthropometric and biochemical indices, epidemiological surveys, and liver ultrasound examinations were collected and analysed statistically, focusing on the association between TG/HDL-C and the incidence of MAFLD.

Results: During a mean follow-up period of 7.62 ± 3.99 years, 24,838 participants developed MAFLD. The cumulative MAFLD incidence rates associated with the first to fourth quartiles of TG/HDL-C were 59.16%, 65.04%, 71.27%, and 79.28%, respectively. The multivariate Cox proportional hazards regression model revealed that the hazard ratios (HRs) (95% CIs) for MAFLD in the second, third, and fourth quartiles were 1.20 (1.16-1.25), 1.50 (1.45-1.56), and 2.02 (1.95-2.10) (P for trend < 0.05), respectively, and the HR (95% CI) corresponding to an increase of one standard deviation in TG/HDL-C was 1.10 (1.09-1.11) (P < 0.05). Subsequent subgroup and sensitivity analyses yielded results similar to those of the main analyses.

Conclusions: TG/HDL-C is independently associated with MAFLD risk, with higher TG/HDL-C indicating greater MAFLD risk.

Objective: The aim of our study was to evaluate the indocyanine green (ICG) retention test as a noninvasive marker of esophageal varices(EV).

Methods: The clinical data of patients diagnosed with compensated liver cirrhosis in Tianjin Second People's Hospital between January 2018 and January 2021 were analysed with SPSS 23.0.

Result: A total of 144 patients (88 M/56 F, 51.7 ± 11.06 years) were enrolled. The ICG retention at 15 min(ICG-r15), PVD, TBIL, Cholinesterase(CHE), AST to ALT ratio(ARR), APRI, splenic area, Lok index, Park index and liver stiffness measurement in the absent or small EV group were lower than those in the medium or large EV group, while the ICG disappareance rate(ICG-K), Effective hepatic blood flow(EHBF), ALB, PLT, and Platelet to Spleen Diameter Ratio(PSDR) were higher, and the differences were significant (P < 0.05). ICG-r15, splenic area, APRI and PLT were independent predictors for medium or large esophageal varices (OR = 1.115, 1.025, 0.281, and 0.987, respectively,P < 0.05). The predictive value of ICG-r15 for medium or large varices was 17.95%, the specificity was 0.849, and the sensitivity was 0.662, the AUROC was 0.815. The cut-off value of PLT for M/L EV was 113.5, and the specificity and sensitivity were 0.616 and 0.887, the AUROC was 0.759. The AUROC of ICG-r15 combined with PLT was 0.866, which was more superior than others.

Conclusion: Although we are far from the replacement of endoscopy, ICG-r15 combined with PLT seems to be able to identify patients with medium or large EV in patients with compensated liver cirrhosis.

Background: Deep learning has made significant advancements in the field of digital pathology, and the integration of multiple models has further improved accuracy. In this study, we aimed to construct a combined prognostic model using deep learning-extracted features from digital pathology images of pancreatic ductal adenocarcinoma (PDAC) alongside clinical predictive indicators and to explore its prognostic value.

Methods: A retrospective analysis was conducted on 142 postoperative pathologically confirmed PDAC cases. These cases were divided into training (n = 114) and testing sets (n = 28) at an 8:2 ratio. Tumor whole-slide imaging features were extracted and screened to construct a pathological risk model based on a pre-trained deep learning model. Clinical and pathological data from the training set were used to select independent predictive factors for PDAC and establish a clinical risk model using LASSO, univariate, and multivariate Cox regression analyses. Based on the pathological and clinical risk models, a combined model was developed. The Harrell concordance index (C-index) was computed to assess the predictive performance of each model for PDAC survival prognosis.

Results: For the training and testing sets, the C-index values for the clinical risk model were 0.76 and 0.75, respectively; for the pathological risk model, they were 0.82 and 0.73, respectively; and for the combined model, they were 0.86 and 0.77, respectively. The combined model exhibited appropriate calibration at 1-, 3-, and 5-year time points, as well as a superior area under the curve of the receiver operating characteristic curve and clinical net benefit compared to the single models.

Conclusions: Integrating the pathological and clinical risk models may provide a higher predictive value for survival prognosis.

Background: The presence of API2/MALT1 fusion in gastric mucosa-associated lymphoid tissue (MALT) lymphoma predicts poor response to Helicobacter pylori (Hp) eradication therapy. This study aimed to assess the correlation between endoscopic morphology of MALT lymphoma and API2/MALT1 fusion and evaluate treatment response to Hp eradication based on morphological subtypes.

Methods: A retrospective review was conducted on patients diagnosed with gastric MALT lymphoma between January 2011 and December 2022. Endoscopic morphology was categorized as superficial, non-superficial, or mixed type. The superficial type was further classified into gastritis superficial lesion and localized superficial lesion based on border clarity. Logistic regression models evaluated the impact of clinical and endoscopic characteristics on anti-Hp therapy effectiveness.

Results: Among the 114 patients included, 93 (81.6%) were Hp-positive, and API2/MALT1 fusion was detected in 58 (50.9%) cases, The superficial type was the predominate morphology (73/114, 64%). The regular arrangement of collecting venules (RAC) sign was noted in 21 (18.4%) cases. In superficial subtypes, the RAC signs were more frequently observed in localized lesion than gastritis lesion (35.6% vs. 7.1%, p = 0.01). and the superficial localized lesion was more common in individuals with positive API/MALT1 fusion than negative ones (76.9% vs. 44.1%, p = 0.01). Following Hp eradication, the remission rate for localized lesion was 34.3%, significantly lower than for gastritis lesion (66.7%, p = 0.01). Both endoscopic morphology (OR = 0.26, 95% CI 0.09-0.75) and API2-MALT1 fusion (OR = 14.29, 95% CI 4.19-48.67) impacted the efficacy of anti-Hp therapy. However, multivariate analysis identified API2-MALT1 fusion as the only independent predictor of treatment outcome (OR = 12.18, 95% CI 3.49-42.55, p < 0.001).

Conclusion: Gastric MALT lymphomas with superficial-type morphology, particularly those with defined borders resembling early gastric cancer, were associated with API2/MALT1 fusion and a lower remission rate after Hp eradication therapy. This suggests that endoscopic morphology, along with API2/MALT1 fusion status, could help predict the therapeutic response, with API2/MALT1 fusion serving as a critical indicator of treatment resistance.

Background: Non-invasive measurement of liver stiffness (LS), traditionally performed in the supine position, has been established to assess liver fibrosis. However, fibrosis degree is not the sole determinant of LS, necessitating the identification of relevant confounders. One often-overlooked factor is body posture, and it remains unclear whether normal daily postures interfere with LS irrespective of fibrosis. A prospective two-group comparison study was conducted to investigate the relationship between posture and LS.

Methods: Sixty-two adults participated, divided into two groups: patients with chronic liver disease and healthy controls. Both groups were assessed using transient elastography (TE) under the supine, seated, and standing postures. Randomization was applied to the order of the two upright postures. A two-way mixed ANOVA was conducted to assess the posture-dependence of LS and its variations between two groups.

Results: Results showed that posture differentially affected LS depending on the presence of liver fibrosis. In 31 healthy individuals (baseline LS range: 3.5-6.8 kPa), a transition from the supine (5.0 ± 1.0 kPa) to seated (5.7 ± 1.4 kPa; p = 0.036) or standing (6.2 ± 1.7 kPa; p = 0.002) positions increased LS, indicating liver stiffening. Conversely, in 31 patients with varying fibrosis stages (baseline LS range: 8.8-38.2 kPa), posture decreased LS from the supine (15.9 ± 7.3 kPa) to seated (13.8 ± 6.2 kPa; p < 0.001) or standing (13.9 ± 6.2 kPa; p = 0.001) positions. No significant difference in LS was observed between the seated and standing positions in both groups (control group: 5.7 vs. 6.2 kPa, p = 0.305; patient group: 13.8 vs. 13.9 kPa, p = 1). Additionally, different postures did not elicit significant changes in the success rate (supine, 98.6 ± 4%; seated, 97.6 ± 6%; standing, 99.1 ± 3%; p = 0.258) and IQR/median value (supine, 25 ± 8%; seated, 29 ± 15%; standing, 29 ± 12%; p = 0.117), implying no impact on both measurement feasibility and reliability.

Conclusions: We demonstrated, for the first time, the feasibility of utilizing upright postures as an alternative measurement protocol for TE. We further unravel a previously unrecognized role of transitioning between different postures to assist the diagnosis of cirrhosis. The findings suggested that daily physiological activity of postural changes suffices to alter LS. Therefore, body positioning should be standardized and carefully considered when interpreting LS.

Objective: Ulcerative colitis (UC) and Hashimoto's thyroiditis frequently cooccur in patients with multiple autoimmune conditions, but the specific association between UC and hypothyroidism is unknown. We used Mendelian randomization (MR) methods to determine the causal relationship between UC and hypothyroidism.

Methods: We obtained single nucleotide polymorphisms (SNPs) related to ulcerative colitis (UC) and hypothyroidism from genome-wide association studies (GWAS) available in the public database of the Integrated Epidemiology Unit (IEU). To assess the causal relationship between UC and hypothyroidism, we employed MR-Egger, weighted median, inverse variance weighted (IVW), simple mode, and weighted mode methods. Sensitivity analyses were performed using Cochran's Q test, the horizontal pleiotropy test, and the leave-one-out (LOO) method to assess the reliability of the MR data. The genes corresponding to instrumental variables (IVs) were subjected to Gene Ontology (GO) functional annotation, Kyoto Encyclopedia of the Genome (KEGG) pathway enrichment analysis, and protein-protein interaction (PPI) analysis to explore the mechanisms behind the causal relationships at the gene level.

Results: Forward MR analysis indicated that hypothyroidism was associated with an increased risk of UC (IVW: P = 0.02, OR = 9.71, 95% confidence interval (CI) = 1.36-69.46). In contrast, reverse MR did not demonstrate a causal relationship between UC and hypothyroidism (IVW: P = 0.53). Sensitivity analysis proved the reliability of the results. The PPI network revealed CD247, CD80, and STAT4 as central genes. GO and KEGG analyses revealed significant enrichment of the T cell, gamma interferon (IFN-γ), and programmed cell death-1 (PD-1)/programmed cell death ligand-1 (PD-L1) pathways.

Conclusion: Hypothyroidism was a risk factor for UC. The balance of T-cell differentiation played an important role in the process of hypothyroidism-induced UC, and IL-21 might be the key to finding a cure. Enrichment of PD-1/PD-L1 might attenuate inflammation by suppressing the immune action of T cells.