Introduction

Superficial esophageal squamous cell carcinoma (SESCC) was enabled detection by the innovation of Narrow Band Imaging (NBI). However, NBI has insufficient sensitivity for detecting SESCC when used by inexperienced endoscopists.¹⁾ To solve these problems, the construction and usefulness of endoscopic detection models for SESCC using artificial intelligence (AI) have been reported.²⁾ On the other hand, the assistant effect of the AI model for endoscopists has not been fully examined.

Aims & Methods

The aim of this study is to create an AI model to assist in detecting SESCC and to assess the effect of the AI assistance on the endoscopist's diagnostic performance using endoscopic video. We collected endoscopic videos of cases with lesions and without lesions from May 2020 to September 2021 for training data to create an AI model for detecting SESCC and from October 2021 to January 2022 for test data. All videos were taken 3 to 5 seconds at National Cancer Center Hospital East by using NBI. Eligibility criteria for cases with lesions were as follows: i) histopathologically diagnosed ESCC or high-grade intraepithelial neoplasia (HGIN); ii) the depth of lesion shallower than cMM/SM1. Residual lesions after endoscopic resection for ESCC were excluded. The eligibility criteria for cases without lesions were the absence of ESCC or HGIN by endoscopic diagnosis. Patients with a history of chemotherapy, radiotherapy, chemoradiotherapy, or surgery for ESCC were excluded. The performances of the AI model and 18 endoscopists (8 experts and 10 non-experts) for detecting SESCC were evaluated using the test data. Furthermore, to explore the assistance ability of the AI model, the endoscopists were required to diagnose again on the same test data after 4 weeks, referring to the results predicted by the AI model. The diagnostic performance of the endoscopists with and without the AI assistance was compared to assess the sensitivity, specificity, accuracy, and interobserver agreement. The recognition to the lesions by the AI model was defined to detect with Intersection over the Union > 0.3 for 5 consecutive frames for each video. The results of the endoscopists were presented as averages.

Results

We collected 280 cases (140 with lesions and 140 without lesions) videos as training data and 115 cases (52 with lesions and 63 without lesions) videos as test data. The clinicopathological characteristics of the lesions were as follow: median lesion size was 15mm (range:1-100mm), and the pathological depth of tumor were EP/LPM/MM/SM1/SM2=10/17/0/4/4/4. The sensitivity, specificity, and accuracy of the AI model were 76.0%, 79.4% and 77.2% respectively. With the AI assistance, the sensitivity and accuracy of the endoscopists were significantly improved (without vs with the AI assistance: 57.4% vs 66.5%, P<0.05, 68.6% vs 75.9%, p<0.05), while the specificity were slightly improved (87.0% vs 91.6%, P=0.059). For experts, the sensitivity and accuracy were significantly improved (59.1% vs 70.0%, P<0.05, 72.1% vs 79.3%, p<0.05). For non-experts, the accuracy was significantly improved (65.8% vs 73.3%, P<0.05), while the sensitivity and specificity were slightly improved (56.1% vs 63.7%, P=0.07, 81.9% vs 89.2%, p=0.07). The interobserver agreement for diagnosis among the endoscopists was improved with the AI assistance (κ=0.429 vs κ=0.547).

Conclusion

The assistance of the AI model improved the endoscopist's sensitivity of detection of SESCC without reducing the specificity, regardless of their experience. The AI model would be effective in improving the endoscopist's ability to detect SESCC in practice.

References

1) Ishihara R, Takeuchi Y, Chatani R et al. Prospective evaluation of narrow-band imaging endoscopy for screening of esophageal squamous mucosal high-grade neoplasia in experienced and less experienced endoscopists. Dis Esophagus 2010; 23: 480–6.
2) Waki K, Ishihara R, Kato Y et al. Usefulness of an artificial intelligence system for the detection of esophageal squamous cell carcinoma evaluated with videos simulating overlooking situation. Digestive Endoscopy 2021; 33: 1101–1109.

Introduction

In recent years, evidence shows that long non-coding RNAs (lncRNAs) are key regulators of gene transcription and play important roles in the pathogenesis of inflammatory bowel diseases. LncRNAs are involved in the regulation of intestinal epithelial cell apoptosis, cell-cell interactions, and enhancing inflammation, among others. Biological therapies, which are considered the most potent for disease control, only benefit one-third of patients. For this reason, a deeper understanding of the mechanisms by which biological drugs elicit their effect on intestinal mucosal is needed. Hence, we aimed to unravel the ex vivo modulator effect of infliximab on the lncRNAs expression in intestinal biopsies from patients with Crohn´s Disease (CD).

Aims & Methods

We performed an unbiased transcriptomic analysis of intestinal biopsies from the ileum and colon from 30 patients [active CD = 10, quiescent CD = 10, healthy controls (HC) = 10] to identify lncRNA differentially expressed in the setting of infliximab modulation. Endoscopic biopsies were cultured with or without infliximab and the transcriptome was determined by Illumina gene expression array. We used different databases (Ensembl Biomart, RNAcentral and ToppGene) to search for non-annotated lncRNA information and data on the location (cellular component), biological process and molecular function of differentially expressed lncRNAs.

Results

Transcriptomic results revealed a widespread dysregulation of lncRNAs in ileum biopsies from patients with active CD, quiescent CD and HC compared to the colon at baseline and after infliximab culture. These differentially expressed lncRNAs were enriched in such pathways as proliferation, apoptosis, migration, inflammatory response of fibroblasts, response to wounding, posttranscriptional regulation of inflammatory genes and activation of the mitogen-activated protein kinase signaling pathways. Regarding the effect of infliximab according to intestinal location, presence of disease and activity, no significant lncRNAs were identified in the different study comparatives.

Conclusion

We have characterized the basal transcriptomic landscape of lncRNAs in patients with CD (active and quiescent) and HC both in ileum and left colon. However, we have not found differential lncRNA expression due to the effect of infliximab, suggesting that the location (ileum or colon) is more relevant when analyzing differences in lncRNA expression in the intestinal tissue.

Introduction

Patients with immune mediated inflammatory diseases (IMIDs) on immunosuppressive therapies have attenuated vaccine responses (1-3) and are prone to severe infections. Knowledge of COVID-19 outcome following vaccine and hybrid immunity, and identification of protective anti-Spike antibody concentrations are important to further develop vaccine strategies in this vulnerable patient group.

Aims & Methods

In this prospective observational study, adult patients with IMID on immunosuppressive therapies were followed from February 15, 2021, to February 15, 2023. Throughout the study, patients and controls reported data regarding COVID-19, COVID-19 related hospital admissions were recorded from The Norwegian Patient Registry. Anti-Spike antibodies were assessed 2-4 weeks following vaccination and COVID-19. The objectives of this study were to investigate the outcomes of COVID-19 in a large cohort of IMID patients compared to healthy controls, and to identify any associations between anti-Spike antibody concentrations and disease course.

Results

Of 2250 IMID patients included in the study, a total of 1729 with a documented COVID-19 history were included in the present analyses (305 Crohn’s disease, 215 ulcerative colitis, 648 rheumatoid arthritis, 272 psoriatic arthritis, and 289 spondyloarthritis) (median age 54 years (IQR 42-64), 971 (56 %) females). Of the patients, 1080 of 1729 (62%) used anti-TNF treatment (65% monotherapy, 35% combination therapy).
1140 (66%) of patients and 236 of 350 (67%) healthy controls (HC) reported COVID-19, the majority (85%) within the Omicron era. COVID-19 reinfection was reported in in 141 (12%) patients and 32 (14%) HC (p=0.66). Compared to hybrid immunity (COVID-19 after three vaccine doses), the risk of COVID-19 was six times higher following vaccine series only (HR 5.9, (95% CI [4.45, 7.80]), p<0.001). Anti-Spike antibody concentrations <8000 BAU/ml were predictive of COVID-19 after three (HR 1.5 (95% CI [1.14, 1.90]), p=0.003) and four vaccine doses (HR 1.4 (95% CI [1.13, 1.83]), p=0.003), and in hybrid immunity (HR 2.6 (95% CI [1.32, 5.13]), p=0.006). In the entire cohort, the median anti-Spike antibody concentration after the 2^nd vaccine dose was 2823 BAU/ml (IQR 999-6199), after the 3^rd dose 6529 BAU/ml (IQR 2325-10140), after the 4^th dose 7624 BAU/ml (3809-12751) and following hybrid immunity 23506 BAU/ml (11423, 37007). Hospitalisation due to COVID-19 (severe disease) occurred in 22 (2%) patients, 9 (41%) before any vaccination, and none of HC. Four patients were admitted to intensive care. Prior to hospitalisation, the median anti-Spike antibody concentration was 444 BAU/ml (IQR 31-1634). Patients with severe disease were older than non-hospitalised (median age 61 years (IQR 48-74) vs. 54 (42-64), p=0.04) and had a higher frequency of comorbidities (19/22 (86%) vs. 474/1118 (42%), p=0.02). No COVID-19 related deaths occurred.
Prolonged COVID-19 (symptoms >14 days) were reported by 201 (18%) patients compared to 29 (12%) HC (p=0.09). The risk of prolonged disease was higher in the vaccine group compared to the hybrid group (HR 8.5 (95 % CI [3.84, 18.90]), p<0.001).

Conclusion

IMID patients and healthy controls had a comparable occurrence of COVID-19, prolonged disease and reinfection. However, severe disease developed only in the patient group. Hybrid immunity protects against COVID-19 and prolonged disease. A high anti-Spike antibody concentration protects against COVID-19, supporting the role of repeated vaccination in IMID patients on immune-suppressive therapy.

References

1. Jorgensen KK, Hoivik ML, Chopra A, Benth JS, Ricanek P, Moum PB, et al. Humoral immune response to SARS-CoV-2 vaccination in patients with inflammatory bowel disease on immunosuppressive medication: association to serum drug levels and disease type. Scand J Gastroenterol. 2023:1-9.
2. Bjorlykke KH, Orbo HS, Tveter AT, Jyssum I, Sexton J, Tran TT, et al. Four SARS-CoV-2 vaccine doses or hybrid immunity in patients on immunosuppressive therapies: a Norwegian cohort study. Lancet Rheumatol. 2023;5(1):e36-e46.
3. Syversen SW, Jyssum I, Tveter AT, Tran TT, Sexton J, Provan SA, et al. Immunogenicity and Safety of Standard and Third-Dose SARS-CoV-2 Vaccination in Patients Receiving Immunosuppressive Therapy. Arthritis Rheumatol. 2022;74(8):1321-32.

Disclosure

Have received speakers bureaus from Bristol-Myers Squibb and Roche

Introduction

The need for endoscopic necrosectomy (EN) has increased dramatically, as endoscopic EUS-guided drainage has become gold standard for treatment of pancreatic walled-off-necrosis (WON)[1,2]. Although snares and baskets are commonly used in EN procedures, they can be ineffective and dedicated instruments specifically designed for EN are scarce. Recently, an innovative powered endoscopic debridement system (EndoRotor, Interscope Inc., Northbridge, MA, USA) designed to excise and remove necrotic tissue within the WON was introduced specifically for EN with promising results[3,4]. While the first generation EndoRotor catheter was limited to 3.2 mm diameter, a larger catheter that fits into a large-channel gastroscope with a working channel of 6.0 mm has now been introduced. This study evaluates the clinical outcomes, safety, and feasibility of the novel 6.0 catheter for EN of WON.

Aims & Methods

This retrospective study was performed using prospectively collected data from all patients who underwent EN with the 6.0 catheter between December 2021 and March 2023 at Gastro Unit, Copenhagen University Hospital Hvidovre, a tertiary referral center in Denmark. All patients with symptomatic WON and need for necrosectomy were considered eligible for this study. Technical feasibility, safety, and clinical outcomes were evaluated.

Results

Sixteen patients underwent a total of 25 EN procedures with the 6.0 catheter. The mean age of the patients was 51.8 years (SD 17.7 years) and 11 (68.8%) were male. The etiology of the acute necrotizing pancreatitis was alcohol in 4 patients (25.0%), gall stones in 7 patients (43.8%), post-endoscopic retrograde cholangiopancreatography (ERCP) in 3 patients (18.8%), and idiopathic in 2 patients (12.5%). The mean size of the WON was 27.5 cm (SD 10.5 cm). The median computed tomography severity index (CTSI) was 8.5 (P25-P75: 6.0-9.0) and median modified CTSI (mCTSI) was 9.0 (P25-P75: 8.0-10). Two patients did not achieve clinical resolution (12.5%), as they died from multisystem organ failure before resolution of the WON was achieved. The median length of stay in hospital for the patients who achieved clinical resolution was 56.0 days (P25-P75: 36.5-96.0 days) and the median number of necrosectomy procedures performed to achieve clinical resolution was 5.0 (P25-P75: 2.0-8.5). All patients had EN procedures performed with snares in addition to the procedures with the powered debridement system. Five patients had necrosectomy performed with the 3.2 catheter (31.0%) as well as the 6.0 catheter. The median duration of EN with the 6.0 catheter was 60.0 minutes (P25-P75: 45.0-100 minutes) per procedure.
A device-related adverse event was noted in one instance (4.0%). The patient exhibited symptoms of peritoneal reaction in the hours following the procedure, which may have been caused by perforation of the WON into the peritoneal cavity. The patient was treated with laparoscopic peritoneal lavage and drainage and recovered without further interventions. No other procedure-related adverse events were identified. Malfunction of the catheter was observed in 7 procedures (28.0%).

Conclusion

To perform necrosectomy procedures optimally, dedicated endoscopic instruments designed for debridement of pancreatic WON are required. This study suggests that EN with the novel EndoRotor 6.0 catheter is safe and feasible with promising clinical results for EN of pancreatic WON. To assess the efficacy relative to standard techniques, comparative studies are warranted.

References

[1] Karstensen JG, Novovic S, Hansen EF, et al. EUS-guided drainage of large walled-off pancreatic necroses using plastic versus lumen-apposing metal stents: A single-centre randomised controlled trial. Gut 2022; 0: 1–7. doi:10.1136/gutjnl-2022-328225
[2] Ebrahim M, Werge MP, Hadi A, et al. Clinical outcomes following endoscopic or video-assisted retroperitoneal management of acute pancreatitis with large (>15 cm) walled-off pancreatic necrosis: Retrospective, single tertiary center cohort study. Dig Endosc 2022; 34: 1245–1252. doi:10.1111/den.14295
[3] Stassen PMC, de Jonge PJF, Bruno MJ, et al. Safety and efficacy of a novel resection system for direct endoscopic necrosectomy of walled-off pancreas necrosis: a prospective, international, multicenter trial. Gastrointest Endosc 2022; 95: 471–479. doi:10.1016/J.GIE.2021.09.025
[4] Wiel SE van der, May A, Poley JW, et al. Preliminary report on the safety and utility of a novel automated mechanical endoscopic tissue resection tool for endoscopic necrosectomy: a case series. Endosc Int Open 2020; 08: E274–E280. doi:10.1055/A-1079-5015

Introduction

Endoscopic retrograde cholangiopancreatography (ERCP) with endoscopic sphincterotomy (EST) is an established endoscopic treatment for common bile duct stones and acute cholangitis. Due to its less invasiveness compared to surgery, it has been accepted as a first choice even for the elderly receiving anticoagulants. However, there is only a limited number of reports of the safety of EST after minimal cessation of anticoagulant based upon the Japan Gastroenterological Endoscopy Society (JGES)guideline.

Aims & Methods

To evaluate the safety, we aimed to clarify the bleeding risk in patients taking anticoagulants in our institute. Patients who underwent ERCP at double institutes from January 2018 to December 2021. Among them, we compared the rate of bleeding after EST and bleeding requiring blood transfusion in anticoagulant patients and non-administrators. Patients taking anticoagulants were allowed to discontinue the direct oral anticoagulant (DOAC) on the day before or the day of ERCP, and those who followed the JGES guideline for warfarin were included.

Results

ERCP was performed in 1616 patients and an analysis of a total of 882 patients of EST including 69 patients taking anticoagulants revealed that there were no significant differences in the bleeding rate although the blood transfusion rate was significantly higher in the DOAC group (2/52 3.9%) and antiplatelet drug combination cases (1/15 6.7%) compared to other cases (5/805 0.6%). Univariate and multivariate analyzes were performed according to background age, sex, hypertension, diabetes mellitus, heart disease, cerebrovascular disease, hemodialysis, platelet count decreased (less than 10x10⁴/μL), and urgency (within 24 hours after consultation). In the univariate analysis for EST bleeding, cerebrovascular disease, hemodialysis, and thrombocytopenia were significant risk factors(p<0.05). As a result of the multivariate analysis of these, only cerebrovascular disease was a significant risk factor (OR=2.11, 95%CI: 1.21-3.68, p=0.0086). In the univariate analysis for blood transfusion, DOAC users, cerebrovascular disease, and hemodialysis patients were significant risk factors(p<0.05). A multivariate analysis that added EST bleeding to these results also found significant risk factors for similar items. DOAC(OR=6.89, 95%CI: 1.06-44.65, p=0.043) and cerebrovascular disease(OR=18.85, 95%CI: 3.01-117.93, p=0.0017), hemodialysis(OR=55.68, 95%CI: 6.51-475.99, p=0.00024) were significant risk factors for severe bleeding requiring blood transfusion.

Conclusion

Although there were no significant differences between the bleeding rates of EST for those taking anticoagulants and those not taking anticoagulants, it was suggested that there is a possibility of serious bleeding requiring blood transfusion in patients receiving anticoagulants, especially DOAC.

Disclosure

none