Introduction

Colorectal endoscopic submucosal dissection (ESD) has become a widely used endoscopic technique for superficial colorectal tumors resection. ESD has a higher en bloc resection rate particularly for large lesions than endoscopic mucosal resection (EMR). Thus, ESD is associated with a better curability and can facilitate an adequate histological evaluation. However, the high incidence of postoperative adverse events, such as delayed bleeding, delayed perforation, and post-ESD electrocoagulation syndrome (PECS), which are associated with larger mucosal defects after ESD, has been a cause of concern. Recent meta-analyses have reported that the incidence of postoperative adverse events after ESD was higher than that after EMR.¹ Therefore, these events should be controlled. Various closure techniques have been used to close larger mucosal defects in ESD.² However, the efficacy of these closure techniques in preventing adverse events remains controversial. Furthermore, few studies have been conducted to evaluate a single closure technique with a substantial number of cases.
In some cases, closure of mucosal defects after colorectal ESD is performed using a double-layered suturing technique with conventional hemostatic clips.³ In this technique, the muscle layer is first grasped directly with clips and tented inward, and then the mucosal layer is closed. Therefore, this technique has the potential to provide more reliable closure and prevent the spread of inflammation and abdominal pain than methods that merely close the mucosal layer.

Aims & Methods

This study retrospectively evaluated the impact of closure using a double-layered suturing technique on postoperative adverse events.
In total, 370 lesions in 317 patients who underwent colorectal ESD were included in this analysis. Then, patients with 197 lesions that were completely closed were assigned to the closure group. Meanwhile, patients with 173 lesions including 55 that were partially closed and 118 that were not closed were assigned to the non-closure group. Propensity score matching was performed and 136 lesions were selected for each group.

Results

The closure group had a significantly lower overall incidence rate of postoperative adverse events including delayed bleeding, delayed perforation and post-ESD electrocoagulation syndrome than the non-closure group (2.2% vs. 9.6%, p = 0.018). The closure group had a significantly lower incidence of abdominal pain on the day after ESD than the non-closure group (2.9% vs. 11.0%, p = 0.015).

Conclusion

Prophylactic closure of mucosal defects after colorectal ESD using a double-layered suturing technique could prevent postoperative adverse events and abdominal pain on the day after ESD.

References

1 Wang N, Shu L, Liu S, et al. Comparing endoscopic mucosal resection with endoscopic submucosal dissection in colorectal adenoma and tumors: Meta-analysis and system review. PLoS One 2023;18:e0291916.
2 Kobara H, Tada N, Fujihara S, Nishiyama N, Masaki T. Clinical and technical outcomes of endoscopic closure of postendoscopic submucosal dissection defects: Literature review over one decade. Dig Endosc 2023;35:216-231.
3 Tanaka S, Toyonaga T, Obata D, Ishida T, Morita Y, Azuma T. Endoscopic double-layered suturing: a novel technique for closure of large mucosal defects after endoscopic mucosal resection (EMR) or endoscopic submucosal dissection (ESD). Endoscopy 2012;44 Suppl 2 UCTN:E153-154.

Introduction

The performance of large language model (LLM) across specific medical domains is uncertain. We evaluated the performance of ChatGPT-4 and two Chinese LLMs in answering open-ended questions regarding gastroenterology.

Aims & Methods

We selected ChatGPT (4-Turbo, OpenAI) and two Chinese LLMs (model A and B), and embedded our own digestive disease-related knowledge database in the LLMs via retrieval-augmented generation (RAG). We presented 200 questions, with 3 different expressions for each question and 5 repeated queries for each expression. Questions were selected from those commonly encounter in a clinical setting from patients. The accuracy and readability were assessed using a three-classification criterion (good, borderline, poor). Responses with “good” rating in accuracy were further assessed using a 3-point scale. For responses with “poor” rating in accuracy, the LLMs were further prompted to self-correct. The stability was assessed through the consistency in changing-expression and repeating-query groups.

Results

For accuracy evaluation, model B demonstrated superior accuracy, with 94.9% of “good” rating, compared to 92.3% in model A and 90.3% in ChatGPT-4 (all P <0.001). Three LLMs showed comparable average scores in “good” rating responses (2.590 vs 2.573 vs 2.536, P=0.754). For readability evaluation, ChatGPT-4 and model A demonstrated superior readability, with 94.4% and 93.1% of “good” rating, compared to 76.3% in model B (all P< 0.001). The average response length of ChatGPT-4 was comparable to model A and significantly shorter than model B (207.45 vs 271.33 vs 336.87, P<0.001). For stability evaluation, model A demonstrated superior stability in repeating queries than ChatGPT-4 and model B. The stability in changing different expressions was not satisfactory for all the three LLMs and needs to be further improved.

Conclusion

Our study preliminarily demonstrated three LLMs' potential in accurately answering questions regarding gastroenterology, and their application to clinical practice still needs further exploration.

Disclosure

Authors declare no competing interest for this article.

Introduction

Sepsis is a life-threatening condition characterized by organ dysfunction resulting from a dysregulated host response to infection. A defining feature of sepsis is the presence of bacteria in the bloodstream or the translocation of intestinal microbes to distant organs such as the lungs. Many of these pathogens originate from the intestinal microbiota. The intestinal mucosa serves as a critical barrier, and its damage or dysfunction is considered a central factor enabling bacterial invasion and systemic spread during sepsis.

Aims & Methods

This study aims to characterize sepsis-induced dysfunction of the intestinal barrier. Specifically, we investigate whether, and how, the functions of the gastrointestinal epithelial barrier are modulated by different sepsis-causing bacterial strains. To this end, an in vitro model was developed using intestinal epithelial cell lines (HCT-116 and Caco-2) co-cultured with bacterial isolates. These isolates, representing diverse sepsis etiologies, were obtained from the blood of septic patients. The impact of bacterial interaction on epithelial integrity was assessed by quantifying the protein levels of E-cadherin and Occludin using Western blot analysis.

Results

Co-culture with sepsis-associated bacterial strains led to the disruption of epithelial cell-cell junctions. Specifically, Escherichia coli strains BK10478, BK01267, and BK06223, at a multiplicity of infection (MOI) of 5, induced a marked downregulation of both Occludin (a tight junction protein) and E-cadherin (an adherens junction protein) in HCT-116 cells. Notably, only a subset of these strains elicited similar effects in Caco-2 cells. Some bacterial isolates had no detectable impact on junctional protein expression in either cell line, highlighting strain-specific and cell-type-dependent responses.

Conclusion

Sepsis is a heterogeneous syndrome characterized by complex host-pathogen interactions. Our findings reveal that the ability of sepsis-associated E. coli strains to compromise epithelial cell-cell junctions is both strain-specific and dependent on the epithelial cell type. Given the essential role of barrier integrity in preventing bacterial translocation, these results provide novel insights into sepsis pathogenesis and underscore the therapeutic potential of strategies aimed at preserving or restoring epithelial barrier function.

Introduction

Irritable bowel syndrome (IBS) poses a significant healthcare challenge, characterized by chronic gastrointestinal and extraintestinal symptoms impacting individuals' well-being. Treatment preferences may vary among patients from different ethnic groups, such as Arab and Jewish Israelis, necessitating tailored approaches.

Aims & Methods

A bilingual (Hebrew/Arabic) questionnaire grading patients' preferences of treatment goals was developed. It was administered online in Israeli IBS Facebook groups, as well as in two hospital gastroenterology clinics.

Results

The study included 267 IBS patients (91 Arabs and 176 Jews). Demographic analysis revealed a higher proportion of females in both groups, with a significantly greater percentage among Jews compared to Arabs (84% vs. 64.8%, respectively, p<0.001). The median age for Arabs and Jews was for both 32 (interquartile ranges of 26-42 and 24-62, respectively). Arabs exhibited higher rates of mixed-type IBS and constipation, while Jews had a higher prevalence of predominant diarrhea IBS. Arabs reported more bloating, higher rates of IBS-related comorbidities, and medication usage.
When asked to rate the importance of treatment goals, both populations preferred improvement in abdominal pain, bloating, and regular defecation, while assigning lower importance to improving difficulty in mental and/or physical aspects of intercourse, as well as arthralgia and myalgia. Arab patients assigned lower importance scores to various symptoms compared to their Jewish counterparts.

Conclusion

This study highlights the impact of ethnicity on patients' treatment goals. Understanding patient’s preferences will enable tailoring an individual approach to each IBS patient.

Disclosure

None

Introduction

Colorectal cancer (CRC) remains a major cause of mortality and morbidity worldwide. Projected increases in CRC cases pose a significant challenge to the capacity of colonoscopy programs. Referrals for symptomatic colonoscopy exceed capacity in Ireland and most European nations. Fecal Immunochemical Test (FIT) with a lower cut-off (50ng/ml) has been proposed as a tool to aid triage of referrals which do not meet criteria for urgent colonoscopy, potentially allowing colonoscopy to be avoided.

Aims & Methods

The aim of this study was to evaluate FIT testing outcomes when combined with an endoscopy nurse delivered clinical triage and evaluate suitability as a ‘rule out’ test for serious gastrointestinal pathology.
This multi-centre study included symptomatic patients from sixteen hospitals across Ireland who were referred for colonoscopy from 2023-2024. All patients received ‘FIT testing at the front door’. Using an advanced endoscopy nurse triage approach, assessing patient clinical characteristics and FIT result (positive result determined as FIT≥50ng/mL), patients were triaged into urgent colonoscopy or remained on the routine (P2) colonoscopy list.
Among 7,808 patients sent for FIT testing, 408 were excluded on the basis of missing FIT results (n = 407) or incomplete data (n = 1). The remaining 7,400 patients were included in an analysis assessing FIT positivity rates (FIT≥50 ng/mL) as well as factors associated with positive results. A subset of 1,815 patients from five hospitals, who had lower endoscopy outcomes available, were further analysed to determine frequencies of endoscopy findings (ex: normal, CRC, and IBD) and to evaluate the sensitivity, specificity and negative predictive value of FIT-centered endoscopy nurse clinical triage. IBM SPSS Statistics 29.0.1.0 was used for statistical analysis.

Results

Of 7,400 patients who underwent FIT testing across sixteen sites, 16.4% (n=1,215) had a positive result (FIT≥50ng/mL). Multivariate analysis showed that increased age (OR = 1.029; 95% CI = 1.025-1.033; p-value<0.001) and male sex (OR = 1.44; 95% CI = 1.273-1.63; p-value<0.001) were significantly associated with positive FIT results.
At five sites, 1,815 patients who underwent FIT testing were referred for colonoscopy. At these sites, 79.0% (n=1,433) had a FIT<50ng/mL of whom 0.14% (n=2) had CRC and 1.5% (n=21) had significant findings at colonoscopy (ex: CRC or IBD). In contrast, 21.0% (n=382) had a FIT≥50ng/mL of whom 6.8% (n=26) had CRC and 19.1% (n=73) had significant findings at colonoscopy (ex: CRC or IBD).
The proportion of patients with CRC was significantly greater when FIT≥50ng/mL compared to FIT<50ng/mL (Fisher’s Exact Test, p-value <0.001). The sensitivity, specificity, and negative predictive value of FIT≥50ng/mL for CRC at subsequent colonoscopy is 92.9% (95%CI=75.0-98.8%), 80.1% (95%CI=78.1-81.9%), and 99.9% (95%CI=99.4-99.9%), respectively.

Conclusion

FIT testing is useful in low risk symptomatic patients referred for colonoscopy and when combined with enhanced nurse led clinical triage has a very good negative predictive value for serious gastrointestinal pathology.

Introduction

A small-scale study modeled the cost-effectiveness of colorectal cancer (CRC) screening, incorporating productivity factors.

Aims & Methods

Using real-world data, this research aimed to evaluate the cost-utility of Taiwan's CRC screening program, factoring in potential lifetime productivity savings. After interlinkages among databases of the Taiwan Cancer Registry, National Mortality Registry, Taiwan CRC Screening Database, and National Health Insurance claim records, we estimated lifetime survival functions for stage-specific subcohorts using a rolling extrapolation algorithm after stratified by sex and age from 2004 to 2017. We multiplied the survival rates with average utility values of EQ-5D collected from CRC cases at National Cheng Kung University Hospital, to sum up quality-adjusted life expectancy (QALE). We multiplied monthly medical costs and insured salaries by survival probabilities. We summed them over a lifetime for both the index and age-, sex-, and calendar-year matched reference subcohorts for comparison. Lifetime productivity (LP) was assessed only up to age 65. After weighing different stage distributions, we calculated the differences in total QALE loss, lifetime medical expenditures (LME), and LP between screened versus non-screened groups. Finally, we estimated the costs associated with implementing fecal immunochemical testing (FIT) in Taiwan's CRC screening program. We applied a 3% annual discount and an exchange rate of 30.58 New Taiwan Dollars per U.S. Dollar (USD) at the end of 2018.

Results

We enrolled 77,169 CRC patients aged 50–74 in the study. After weighted by different stage distribution, a screen-detected CRC case would save 2.6 (95% CI, 2.3–2.9) and 2.5 (95% CI, 2.1–2.6) QALYs for women and men, respectively, accompanied by 7,000 USD (95% CI, 4,000–10,000) and 8,000 USD (95% CI, 6,000–10,000) in LME, and 17,000 USD (95% CI, 16,000–19,000) and 23,000 USD (95% CI, 22,000–25,000) in LP. For every 100,000 persons screened, the program spent 682,000 USD for FIT, 1,051,000 USD and 1,987,000 USD for men and women, respectively, for colonoscopy, 61,000 USD and 133,000 USD for surveillance colonoscopy and 9,000 USD and 15,000 USD for treating complications. The incremental cost-utility ratio (ICUR) of 6,650 and 6,230 USD per QALY for women and men, respectively, would become 120 and -2970 USD per QALY after accounting for lifetime productivity.

Conclusion

After considering lifetime productivity, Taiwan's CRC screening program would be close to being cost-saving. Future corroboration studies are warranted.

Introduction

Polypectomy is a common endoscopic procedure with cold snare and hot snare techniques being widely used. Hot snare polypectomy involves cautery, which may increase the risk of thermal injury and related complications. Cold snare polypectomy has a well-established safety and efficacy profile for small colonic polyps, however, comparative data on the clinical outcomes of these two methods for gastric polyps remains limited.

Aims & Methods

A retrospective analysis was conducted on 161 consecutive gastric polypectomies for polyps under 10mm, comprising a sample of 41 patients that underwent cold snare polypectomy and 120 patients that underwent hot polypectomy over a period of 5 years.

Results

The mean polyp size differed significantly between groups, being smaller in the cold snare group (5.20 ± 1.18 mm vs 6.34 ± 1.49 mm; p < 0.001). The Charlson Comorbidity Index averaged 3 points, with no significant differences between the groups regarding the use of anticoagulant or antiplatelet therapy. Statistically significant differences were observed in both the location and histological type of the polyps between the two techniques (p = 0.003 ; p < 0.001). Cold snare polypectomy was more frequently used for polyps located in the fundus and for fundic gland polyps, whereas hot snare polypectomy was more commonly utilized for polyps located in the cardia and antrum. Submucosal injection with gelofusine and adrenaline were more frequently associated with the hot snare technique (p = 0.001). Regarding clinical outcomes, intra-procedural bleeding occurred in 16.1% of cases and was successfully managed endoscopically in all instances, mainly with clips (76.9%). No cases of delayed bleeding or perforation were observed. There were no significant differences between the groups concerning complete histological resection or polyp recurrence.

Conclusion

Cold snare polypectomy demonstrated comparable safety and efficacy to hot snare polypectomy in the resection of small gastric polyps, possibly representing a preferable approach to avoid thermal injury and related complications.