Diagnostic performance of deep-learning-based virtual chromoendoscopy in gastric neoplasms.

BACKGROUNDS: Cycle-consistent generative adversarial network (CycleGAN) is a deep neural network model that performs image-to-image translations. We generated virtual indigo carmine (IC) chromoendoscopy images of gastric neoplasms using CycleGAN and compared their diagnostic performance with that of white light endoscopy (WLE). METHODS: WLE and IC images of 176 patients with gastric neoplasms who underwent endoscopic resection were obtained. We used 1,633 images (911 WLE and 722 IC) of 146 cases in the training dataset to develop virtual IC images using CycleGAN. The remaining 30 WLE images were translated into 30 virtual IC images using the trained CycleGAN and used for validation. The lesion borders were evaluated by 118 endoscopists from 22 institutions using the 60 paired virtual IC and WLE images. The lesion area concordance rate and successful whole-lesion diagnosis were compared. RESULTS: The lesion area concordance rate based on the pathological diagnosis in virtual IC was lower than in WLE (44.1% vs. 48.5%, p < 0.01). The successful whole-lesion diagnosis was higher in the virtual IC than in WLE images; however, the difference was insignificant (28.2% vs. 26.4%, p = 0.11). Conversely, subgroup analyses revealed a significantly higher diagnosis in virtual IC than in WLE for depressed morphology (41.9% vs. 36.9%, p = 0.02), differentiated histology (27.6% vs. 24.8%, p = 0.02), smaller lesion size (42.3% vs. 38.3%, p = 0.01), and assessed by expert endoscopists (27.3% vs. 23.6%, p = 0.03). CONCLUSIONS: The diagnostic ability of virtual IC was higher for some lesions, but not completely superior to that of WLE. Adjustments are required to improve the imaging system's performance.

A mathematical simulation model to determine the optimal endoscopic screening strategy for detection of H. pylori-naïve gastric neoplasms.

BACKGROUND: The effectiveness of esophagogastroduodenoscopy (EGD) screening in cohorts with low Helicobacter pylori prevalence is unknown. This study aimed to develop an optimally efficient EGD screening strategy for detecting H. pylori-naïve gastric neoplasms (HpNGNs). METHODS: EGD data of 12 institutions from 2016 to 2022 were retrospectively analyzed. Age-related HpNGN prevalence, tumor growth rate, missing rate, and detection threshold size were calculated from the databases. Subsequently, using clinical data, a novel mathematical model that simultaneously simulated demographic changes and HpNGN detection was developed. Screening strategies using different starting ages (40/45/50 years) and intervals (2/5/10 years) were also compared. The detection rates of all tumors occurring within the virtual cohort and number-needed-to-test (NNT) were measured as outcomes. RESULTS: Data of 519,368 EGDs and 97 HpNGNs (34 pure signet ring cell carcinomas, 26 gastric adenocarcinomas of the fundic gland type, 30 foveolar gastric adenoma-Raspberry type, and seven undifferentiated-type cancer cases) were analyzed. A virtual cohort with a 70-year time horizon was used to simulate the occurrence, growth, and detection of 346,5836 people. Among the strategies with detection rate > 50%, the screening strategy with a 5-year interval starting at 45 years of age had the lowest NNT. Adopting this strategy, most HpNGNs were detected at < 20 mm in size, and the deep submucosal invasion rate was less than 30%. CONCLUSIONS: A mathematical simulation model revealed that screening every 5 years starting at 45 years of age could efficiently assist in identifying HpNGNs at an early stage.

The Management of Direct Oral Anticoagulants for Colorectal Endoscopic Submucosal Dissection: A Multicenter Observational Study.

BACKGROUND: There is a lack of reports on the use of direct oral anticoagulants (DOACs) during colorectal endoscopic submucosal dissection (ESD). AIMS: We aimed to assess whether the use of DOACs is associated with a higher incidence of delayed bleeding (DB) after ESD. METHODS: A total of 4175 colorectal neoplasms in 3515 patients were dissected at our hospitals during study period. We included 3909 lesions in the final analysis. The lesions were divided into two groups: the no-AT group (3668 neoplasms) and the DOAC group (241 neoplasms). We also compared the DOAC withdrawal group (154 neoplasms) and the DOAC continuation group (87 neoplasms). RESULTS: Among the 3909 lesions, DB occurred in a total of 90 cases (2.3%). The rate of DB was 2.2% (82/3668), and 3.3% (8/241), respectively. There were no significant differences in the rate of DB between the no-AT group and the DOAC group. In the DOAC group, there were no significant differences in the rate of DB between the withdrawal group (5.2%, 8/154) and the continuation group (0%, 0/87). The multivariable analysis identified the location of the lesion in the rectum (odds ratio [OR], 4.04; 95% confidence interval [CI], 2.614-6.242; p < 0.001) and lesions ≥ 30 mm in diameter (OR, 4.14; 95% CI, 2.349-7.34; p < 0.001) as independent risk factors for DB. CONCLUSIONS: Our findings suggest that DOAC use has no significant important on the rate of DB. Prospective studies are warranted to determine whether treatment with DOACs should be interrupted prior to colorectal ESD.

Endoscopic resection for gastric submucosal tumors: Japanese multicenter retrospective study.

OBJECTIVES: Limited information is available on the efficacy and safety of endoscopic resection (ER) of gastric submucosal tumors (SMTs) in Japanese endoscopic practice where ER for gastric tumors is extremely popular. METHODS: We conducted this study to elucidate the current scenario of ER for gastric SMTs in Japanese endoscopic practice. Patients (from 12 institutions) with gastric SMTs who underwent ER were enrolled from the first case until August 2020. RESULTS: We enrolled 117 patients with 118 lesions. The number of patients who underwent ER increased over the years. The mean endoscopic tumor size was 20 ± 7.2 (8-40) mm. The growth type was primarily intraluminal (90%). The mean resection and wound closure times were 58 ± 38 (range, 12-254) min and 31 ± 41 (range, 3-330) min, respectively. Complete ER was achieved for 117 (99%) lesions. Full-thickness resection rate was 44%; however, only 12 (10%) patients required abdominal paracentesis for decompression. Endoscopic treatments were completed in 115 (97%) lesions, while three lesions required conversions to laparoscopic surgery due to luminal collapse, uncontrolled bleeding, and difficulty in defect closure. Gastrointestinal stromal tumors were the most common pathology (74%). No recurrence was observed during the mean follow-up period of 4.3 ± 2.9 years. The 5-year overall survival rate was 98.9% (95% confidence interval 97.8-100%). CONCLUSION: Performance of ER for gastric SMTs is increasing in Japan. The technique seems feasible in Japanese endoscopic practice, warranting further validation in a prospective study.

Safety and Efficacy of Cold Snare Polypectomy Without Submucosal Injection for Large Sessile Serrated Lesions: A Prospective Study.

BACKGROUND & AIMS: Cold snare polypectomy (CSP) has become the standard resection method for small colorectal polyps (<10 mm). Sessile serrated lesions (SSL) have low prevalence of advanced histology irrespective of size, and thus could be amenable to CSP. In this study, we evaluated the safety and efficacy of CSP for SSLs ≥10 mm. METHODS: Between November 2018 and January 2020, we prospectively enrolled 300 consecutive patients who underwent CSP for 474 SSLs ≥10 mm. To delineate SSL borders, indigo carmine chromoendoscopy and/or image-enhanced endoscopy was conducted. Piecemeal CSP (pCSP) was performed in cases where en-bloc resection was difficult. Biopsy specimens were obtained from the margins of the post-polypectomy defect to confirm complete resection. Surveillance colonoscopy was performed to screen for local recurrence. RESULTS: All lesions were successfully resected using CSP without submucosal injection. The median diameter of the resected lesions was 14 mm, and pCSP was used to resect 106 (22%) lesions. Post-polypectomy biopsies revealed residual serrated tissue in only one case (0.2%). Adverse events included immediate bleeding in 8 (3%) patients; no delayed bleeding events occurred, irrespective of the use of antithrombotic drugs. During a 7-month median follow-up period, surveillance colonoscopies were performed for 384 lesions (81%), and no local recurrences were detected. CONCLUSIONS: CSP without submucosal injection is a safe and effective treatment for SSLs ≥10 mm. UMIN Clinical Trials, Number: UMIN000034763.

The feasibility and safety of endoscopic submucosal dissection of gastric lesions larger than 5 cm.

BACKGROUND: As the indications for endoscopic submucosal dissection (ESD) for early gastric cancer have been revised, diagnostic ESD has increased. However, despite the technical difficulty of ESD for large lesions, the degree to which curative resection can be achieved has not been clarified. This study investigated the feasibility and safety of ESD for gastric lesions larger than 5 cm. METHODS: This retrospective multicenter study included 3474 gastric lesions treated by ESD from April 2012 to December 2021. We compared clinicopathological characteristics and treatment outcomes between lesions ≥ 5 cm and lesions < 5 cm. RESULTS: There were 128 lesions in the ≥ 5 cm group and 3282 lesions in the < 5 cm group. In the ≥ 5 cm group, upper location and fibrosis during ESD were more common, with a lower rate of 0-IIc type. Both en bloc resection rate and R0 resection rate were comparable, but there was a difference in curative resection rate (65.6% in the ≥ 5 cm group and 91.5% in the < 5 cm group). The frequency of adverse events (post-ESD bleeding, perforation, or stenosis) was almost similar, but delayed perforation was significantly more common (1.6% in the ≥ 5 cm group vs. 0.1% in the < 5 cm group). CONCLUSIONS: About two-thirds of curative resections were obtained with ESD for early gastric lesions larger than 5 cm, but delayed complications should be noted (Number: UMIN000047725).

Evaluation of colorectal endoscopic submucosal dissection using a multifunctional snare: a prospective clinical feasibility study (with videos).

BACKGROUND AND AIMS: The number of colorectal endoscopic submucosal dissections (ESDs) for early colorectal cancer is expected to increase in the future; therefore, cost reduction is a clinically important issue. The SOUTEN snare (Kaneka Medics, Tokyo, Japan) is a novel multifunctional snare developed for hybrid ESD at a low price. If ESD can be performed safely using the SOUTEN snare, the same therapeutic effect can be obtained as with conventional ESD at a lower cost. The aim of this prospective, pilot, clinical feasibility study was to evaluate the safety and efficacy of ESD using the SOUTEN snare (SOUTEN-ESD). METHODS: From October 2018 to January 2019, 119 consecutive patients (121 ESD procedures, 137 colorectal neoplasms) were prospectively enrolled and treated by SOUTEN-ESD at NTT Medical Center Tokyo and Omori Red Cross Hospital. The outcomes of SOUTEN-ESD were evaluated. RESULTS: Among 137 neoplasms, SOUTEN-ESD was completed in all cases. No cases required conversion to hybrid ESD or to a dedicated ESD device. The mean procedure time was 26.1 ± 14.3 minutes. Both the en-bloc resection rate and R0 resection rate were 100%. The rate of perforation was 0%, the rate of delayed bleeding was 2.2%, and the rate of post-ESD coagulation syndrome was 2.9%. CONCLUSIONS: SOUTEN-ESD was safe and had good outcomes. Although further studies are required to examine indications for SOUTEN-ESD and confirm the results of this study, effective ESD with this novel knife is feasible. The SOUTEN snare is a realistic option for colorectal ESD. (Clinical trial registration number: UMIN 000034299.).

A new classification for the diagnosis of superficial non-ampullary duodenal epithelial tumors using endocytoscopy: A prospective study.

BACKGROUND AND AIM: Although the frequency of endoscopic diagnosis of superficial non-ampullary duodenal epithelial tumors (SNADETs) has been increasing in recent years, no criteria for the endoscopic diagnosis of these tumors have been established yet. The aim of this study was to assess the usefulness of endocytoscopy for diagnosis SNADETs and to establish new criteria. METHODS: This prospective study was conducted at the NTT Medical Center Tokyo from May 2019 to July 2020, and a total of 100 consecutive SNADETs were enrolled. All the endocytoscopic images of the lesions and surrounding normal mucosa were classified into three groups according to the degree of structural atypia and the nuclear morphology and size. The endocytoscopic diagnoses using endocytoscopic classification was compared with the final histopathological diagnoses. RESULTS: Data of 93 patients with 98 lesions were included in the analysis. The preoperative diagnosis by endocytoscopy coincided with the final histopathological diagnosis in 85 (86.7%) of 98 SNADETs. In addition, the sensitivity and specificity for VCL 4/5 were 87.7% and 85.4%, respectively. In contrast, the accuracy, sensitivity, and specificity of preoperative diagnosis by biopsy were 64.3%, 50.9%, and 82.9%, respectively. Preoperative diagnosis by endocytoscopy showed significantly superior accuracy and sensitivity as compared with preoperative biopsy diagnosis (P < 0.001, respectively). CONCLUSIONS: This new classification (endocytoscopic classification) allows prediction of the tumor histopathology in real time, during endocytoscopy without biopsy, and is expected to be of help in determining the appropriate therapeutic strategies for individual cases of SNADETs. (Clinical trial registration number: UMIN000038643.).

Endoscopic submucosal dissection for colorectal neoplasms in proximity or extending to a diverticulum.

BACKGROUND AND AIMS: At specialized facilities, endoscopic submucosal dissection (ESD) has currently been performed even for difficult cases such as tumors extending to a diverticulum that previously required surgery. This study aims to classify the type of lesion according to the degree of infiltration to a diverticulum and assessed the safety and efficacy of ESD for each type of lesion. METHODS: We retrospectively reviewed ESD for lesions at NTT Medical Center Tokyo between January 2014 and April 2019. Lesions were classified as follows: Type 1: lesions in contact with or within 3 mm of the edge of a diverticulum; Type 2: lesions that partially infiltrated into a diverticulum; and Type 3: lesions that infiltrated into and completely covered the diverticulum. Furthermore, ESD strategies were divided into A and B, which indicates that a lesion was resected separately from the diverticulum and along the entire diverticulum, respectively. The clinicopathological characteristics and clinical outcomes were analyzed according to the strategy. RESULTS: A total of 47 lesions satisfied inclusion criteria (19 Type 1, 12 Type 2, and 16 Type 3 lesions). 19 Type 1 and 8 Type 2 lesions were resected using Strategy A, while 4 Type 2 and 16 Type 3 lesions were resected using Strategy B. En bloc resection was achieved in all cases. In Strategy A, the R0 resection rate was 96.3% and the curative resection rate was 88.9%. On the contrary, in Strategy B, the R0 resection rate was 95.0% and the curative resection rate was 90.0%. In Strategy B, one of the patients developed post-operative bleeding that required endoscopic hemostasis; another patient developed delayed perforation that required emergency surgery. CONCLUSIONS: ESD for colorectal neoplasms in proximity or extending to a diverticulum is challenging, but this procedure can be a safe and effective therapeutic option.

Simple scoring system for the diagnosis of superficial non-ampullary duodenal epithelial tumors.

BACKGROUND AND AIMS: Differentiating superficial non-ampullary duodenal epithelial tumors (SNADETs) that harbor malignant potential is important. We developed a simple scoring system and investigated whether it enables the differentiation of low-grade adenoma and high-grade adenoma/adenocarcinoma. PATIENTS AND METHODS: We retrospectively enrolled 197 consecutive patients with 207 SNADETs who underwent endoscopic resection at NTT Medical Center Tokyo between March 2016 and May 2019. Endoscopic findings were compared between Vienna Classification (VCL) C3 and C4/5 lesions. A multivariate logistic regression analysis was performed to develop a scoring system to identify VCL C4/5 lesions. The efficacy of our scoring system was elucidated among five novice and five expert endoscopists. RESULTS: Of 207 SNADETs, 66 and 141 lesions were pathologically diagnosed as VCL C3 and C4/5. A multivariate logistic regression analysis identified a tumor diameter of 10-19 mm (OR, 3.81; 95% CI, 1.02-14.2; P = 0.04), a tumor diameter ≥20 mm (OR, 95.2; 95% CI, 10.4-871.0; P < 0.001), a red color (OR, 14.5; 95% CI, 3.55-59.6; P < 0.001), the presence of irregular surface pattern (OR, 12.4; 95% CI, 3.00-51.4; P < 0.001), and the presence of irregular vessel pattern (OR, 13.7; 95% CI, 4.03-46.6; P < 0.001) as independent significant predictors of VCL C4/5. Considering these results, we developed a scoring system. Using an appropriate cutoff value, the diagnostic accuracy, sensitivity and specificity were calculated as 92%, 95% and 93%. The average diagnostic accuracy did not differ between novice and expert endoscopists (86% vs 87%, P = 0.76). CONCLUSIONS: Our scoring system was useful for differentiating VCL C3 and C4/5 lesions. UMIN Clinical Trials (No. 000039063).

Risk factors of delayed bleeding after endoscopic resection of superficial non-ampullary duodenal epithelial tumors and prevention by over-the-scope and conventional clipping.

OBJECTIVES: This study was aimed to reveal risk factors for delayed bleeding after endoscopic resection (ER) of superficial non-ampullary duodenal epithelial tumors (SNADETs) and at exploring measures to prevent this complication. METHODS: A total of 235 consecutive patients with 249 SNADETs who had undergone ER were enrolled in this study. They were divided into two groups: OTSC group, consisting of the initial 114 cases in which the defects were closed only using OTSCs; and OTSC-c group, consisting of the later 135 cases in which conventional clips were additionally used to cover the inverted submucosa after post-procedure defect closure using OTSCs. The therapeutic outcomes were then compared between the OTSC and OTSC-c groups. RESULTS: All lesions were successfully resected en-bloc, and the R0 resection rate was 92.4%. The complete defect closure rate was 90.0% and no delayed perforation occurred when successful defect closure was achieved. The rate of delayed bleeding was significantly higher in the OTSC group than in OTSC-c group (11.4% vs. 1.5%, P = 0.001). Multivariate logistic regression analyses revealed that tumor location distal to the ampulla (OR 10.0; 95% CI 1.24-81.0, P = 0.03) and use of a DOAC (OR 8.83; 95% CI 1.13-68.7, P = 0.04) were significant independent predictors of delayed bleeding. Propensity score-matching analysis revealed that additional use of conventional clips was associated with a significantly reduced risk of delayed bleeding (P = 0.003). CONCLUSIONS: Additional use of conventional clips after prophylactic defect closure using OTSCs appears to be useful to reduce the risk of delayed bleeding after ER of SNADETs. UMIN Clinical Trials (No. 000035478).

Efficacy of an over-the-scope clip for preventing adverse events after duodenal endoscopic submucosal dissection: a prospective interventional study.

BACKGROUND: Duodenal endoscopic submucosal dissection (ESD) remains technically challenging, with a high risk of severe adverse events. Because exposure of the duodenal post-ESD mucosal defect to pancreatic juice and bile acid reportedly induces delayed perforation and bleeding, we examined whether defect closure using an over-the-scope clip (OTSC) system was useful for preventing postoperative adverse events. METHODS: From April 2016 to February 2017, a total of 50 consecutive patients with superficial non-ampullary duodenal epithelial tumors (SNADETs) larger than 10 mm, with no more than semi-circumferential spread, were prospectively enrolled in this study. All of the lesions were treated by experienced ESD operators and the post-ESD mucosal defect was closed using OTSCs. RESULTS: All of the SNADETs were completely removed by ESD, with an R0 resection rate of 88.0 %. The mean procedure and closure times were 67.3 ±â€Š58.8 minutes and 9.8 ±â€Š7.2 minutes, respectively. Although complete defect closure was achieved in 94.0 % of the patients (47/50), two patients required surgical conversion. Delayed perforation occurred in only one patient (2.1 %), who did not have successful closure of the defect, as misplacement of the OTSC exposed the muscle layer. Meanwhile, delayed bleeding occurred in three patients (6.3 %); however, the bleeding was easily controlled using endoscopic coagulation. The mean duration of postoperative hospitalization was 5.5 ±â€Š7.2 days. CONCLUSIONS: Prophylactic defect closure using OTSCs may be effective in reducing severe adverse events after duodenal ESD.