The diagnostic ability to classify neoplasias occurring in inflammatory bowel disease by artificial intelligence and endoscopists: A pilot study.

BACKGROUND AND AIM: Although endoscopic resection with careful surveillance instead of total proctocolectomy become to be permitted for visible low-grade dysplasia, it is unclear how accurately endoscopists can differentiate these lesions, as classifying neoplasias occurring in inflammatory bowel disease (IBDN) is exceedingly challenging due to background chronic inflammation. We evaluated a pilot model of an artificial intelligence (AI) system for classifying IBDN and compared it with the endoscopist's ability. METHODS: This study used a deep convolutional neural network, the EfficientNet-B3. Among patients who underwent treatment for IBDN at two hospitals between 2003 and 2021, we selected 862 non-magnified endoscopic images from 99 IBDN lesions and utilized 6 375 352 images that were increased by data augmentation for the development of AI. We evaluated the diagnostic ability of AI using two classifications: the "adenocarcinoma/high-grade dysplasia" and "low-grade dysplasia/sporadic adenoma/normal mucosa" groups. We compared the diagnostic accuracy between AI and endoscopists (three non-experts and four experts) using 186 test set images. RESULTS: The diagnostic ability of the experts/non-experts/AI for the two classifications in the test set images had a sensitivity of 60.5% (95% confidence interval [CI]: 54.5-66.3)/70.5% (95% CI: 63.8-76.6)/72.5% (95% CI: 60.4-82.5), specificity of 88.0% (95% CI: 84.7-90.8)/78.8% (95% CI: 74.3-83.1)/82.9% (95% CI: 74.8-89.2), and accuracy of 77.8% (95% CI: 74.7-80.8)/75.8% (95% CI: 72-79.3)/79.0% (95% CI: 72.5-84.6), respectively. CONCLUSIONS: The diagnostic accuracy of the two classifications of IBDN was higher than that of the experts. Our AI system is valuable enough to contribute to the next generation of clinical practice.

The clinicopathological differences of sporadic non-ampullary duodenal epithelial neoplasm depending on tumor location.

BACKGROUND AND AIM: Although sporadic non-ampullary duodenal adenoma is speculated to be precancerous lesion, the relationship between adenoma and carcinoma remains unclear due to their rarity. Previous studies on sporadic non-ampullary duodenal epithelial neoplasm (SNADEN) have mainly targeted superficial tumors, like adenoma and early carcinoma. The clinicopathological features, including those of advanced carcinoma, remain poorly investigated. We assessed the clinicopathological features of SNADEN, including advanced carcinoma, focusing on tumor location. METHODS: We retrospectively collected the data of 410 patients who had been clinically and pathologically diagnosed with SNADEN at 11 institutions in Japan between June 2002 and March 2014. RESULTS: The SNADEN was mucosal neoplasia and invasive carcinoma in 321 (78.3%) and 89 (21.7%) patients, respectively. The proportion of invasive carcinomas in SNADEN was significantly higher on the oral side of the papilla of Vater (oral-Vater) than on the anal side (anal-Vater) (27.9% vs 14.4%, P < 0.001). Undifferentiated-type carcinoma was significantly more frequent with oral-Vater than anal-Vater (38.7% vs 14.8%, P = 0.026). The recurrence rate of surgically R0 resected locally advanced carcinomas was significantly higher with oral-Vater than anal-Vater (46.4% vs 8.3%, P = 0.021). Furthermore, the relapse-free survival with oral-Vater was significantly shorter than with anal-Vater (hazard ratio: 2.35; 95% confidence interval: 1.09-5.50; P = 0.028). CONCLUSIONS: The clinicopathological features of SNADEN on oral-Vater were different from those on anal-Vater. SNADEN on oral-Vater was more likely to be invasive carcinomas and might behave more aggressively due to biologically higher malignancy than that on anal-Vater.

The characteristics and outcomes of small bowel adenocarcinoma: a multicentre retrospective observational study.

BACKGROUND: Small bowel adenocarcinoma (SBA) is a rare malignancy that accounts for 1-2% of gastrointestinal tumours. We investigated the clinical characteristics, outcomes, and prognostic factors of primary SBA. METHODS: We retrospectively analysed the characteristics and clinical courses of 205 SBA patients from 11 institutions in Japan between June 2002 and August 2013. RESULTS: The primary tumour was in the duodenum and jejunum/ileum in 149 (72.7%) and 56 (27.3%) patients, respectively. Sixty-four patients (43.0%) with duodenal adenocarcinoma were asymptomatic and most cases were detected by oesophagogastroduodenoscopy (EGD), which was not specifically performed for the detection or surveillance of duodenal tumours. In contrast, 47 patients (83.9%) with jejunoileal carcinoma were symptomatic. The 3-year survival rate for stage 0/I, II, III, and IV cancers was 93.4%, 73.1%, 50.9%, and 15.1%, respectively. Multivariate analysis revealed performance status 3-4, high carcinoembryonic antigen, high lactate dehydrogenase (LDH), low albumin, symptomatic at diagnosis, and stage III/IV disease were independent factors for overall survival (OS). Ten patients (18.5%) with stage IV disease were treated with a combination of resection of primary tumour, local treatment of metastasis, and chemotherapy; this group had a median OS of 36.9 months. CONCLUSIONS: Although most SBA patients were diagnosed with symptomatic, advanced stage disease, some patients with duodenal carcinoma were detected in early stage by EGD. High LDH and symptomatic at diagnosis were identified as novel independent prognostic factors for OS. The prognosis of advanced SBA was poor, but combined modality therapy with local treatment of metastasis might prolong patient survival.

Multicenter Prospective Study on the Safety of Upper Gastrointestinal Endoscopic Procedures in Antithrombotic Drug Users.

BACKGROUND: The Japan Gastroenterological Endoscopy Society updated its guidelines for gastroenterological endoscopy in patients undergoing antithrombotic treatment in July 2012. However, the safety of endoscopic procedures in antithrombotic drug users has not been fully investigated. AIMS: To evaluate the safety of upper gastrointestinal endoscopic procedures in antithrombotic drug users. METHODS: From September 2013 to September 2015, patients who were taking antithrombotic drugs and who underwent upper gastrointestinal endoscopic procedures were prospectively enrolled at five hospitals. Incidences of bleeding and thrombosis during endoscopic procedures were evaluated. RESULTS: A total of 270 patients [221 for endoscopic mucosal biopsy and 49 for endoscopic mucosal resection (EMR)/endoscopic submucosal dissection (ESD) were enrolled. The bleeding rate was 0.9% for endoscopic mucosal biopsy and 22% for EMR/ESD, respectively. The bleeding rate after endoscopic mucosal biopsy was not significantly high, even if antithrombotic drugs were continued (0 vs. 1%, P > 0.99), while it was significantly higher among multiple antithrombotic drug users than single drug users (5.9 vs. 0%, P < 0.05). The bleeding rate after EMR/ESD was also higher among multiple antithrombotic drug users than single drug users, but was not significantly different (33 vs. 14%, P = 0.17). Moreover, there were no differences in bleeding rates according to the cessation or continuance of antithrombotic drugs (20 vs. 25%, P = 0.74). There were no thromboembolisms in all cases. CONCLUSIONS: Upper gastrointestinal endoscopic procedures performed under the new guidelines appear acceptable. However, endoscopic procedures among multiple antithrombotic drug users show a greater potential for bleeding.

A multicenter survey of enteroscopy for the diagnosis of intestinal follicular lymphoma.

The importance of enteroscopy examinations to investigate the entire length of the small intestines has been emphasized in follicular lymphoma patients with intestinal involvement. The aim of the present study was to determine the current state of enteroscopy examinations, including the performance rate, and the prevalence of small intestinal lesions in a patient population in Japan. A retrospective multicenter survey of 17 institutions collected the case information of 110 follicular lymphoma patients with gastrointestinal involvement. The results of the enteroscopy examinations were reviewed, and in order to identify potential factors affecting the performance rate of enteroscopy, patient gender, age at lymphoma diagnosis, histopathological grade, clinical stage, the date of the initial diagnosis and the annual volume of enteroscopy at the institution were compared between the patients who underwent one or more enteroscopy procedures and the patients who did not undergo enteroscopy. A total of 34 patients (30.9%) underwent enteroscopy, and 24 of these (70.6%) presented with involvement in the jejunum and/or ileum. It was found that more patients diagnosed in recent years and more patients treated at an ultra-high volume institution (≥101 enteroscopy examinations/year) underwent an enteroscopy. In conclusion, although the prevalence of small intestinal lesions was high (70.6%) in the follicular lymphoma patients presenting with intestinal involvement, the performance rate of enteroscopy was only 30.9%, and thus the majority of the patients have not undergone enteroscopy examinations. Further investigation is required to define the clinical significance of enteroscopy at the initial diagnostic work-up and during the follow-up period of these patients.

Diagnostic role of 18F-fluorodeoxyglucose positron emission tomography for follicular lymphoma with gastrointestinal involvement.

AIM: To investigate the capacity for 18F-fluorodeoxyglucose (18F-FDG) positron emission tomography (PET) to evaluate patients with gastrointestinal lesions of follicular lymphoma. METHODS: This retrospective case series consisted of 41 patients with follicular lymphoma and gastrointestinal involvement who underwent 18F-FDG-PET and endoscopic evaluations at ten different institutions between November 1996 and October 2011. Data for endoscopic, radiological, and biological examinations performed were retrospectively reviewed from clinical records. A semi-quantitative analysis of 18F-FDG uptake was performed for each involved area by calculating the maximum standardized uptake value (SUVmax). Based on the positivity of 18F-FDG uptake in the gastrointestinal lesions analyzed, patients were subdivided into two groups. To identify potential predictive factors for 18F-FDG positivity, these two groups were compared with respect to gender, age at diagnosis of lymphoma, histopathological grade, pattern of follicular dendritic cells, mitotic rate, clinical stage, soluble interleukin-2 receptor levels detected by 18F-FDG-PET, lactate dehydrogenase (LDH) levels, hemoglobin levels, bone marrow involvement, detectability of gastrointestinal lesions by computed tomography (CT) scanning, and follicular lymphoma international prognostic index (FLIPI) risk. RESULTS: Involvement of follicular lymphoma in the stomach, duodenum, jejunum, ileum, cecum, colon, and rectum was identified in 1, 34, 6, 3, 2, 3, and 6 patients, respectively. No patient had esophageal involvement. In total, 19/41 (46.3%) patients exhibited true-positive 18F-FDG uptake in the lesions present in their gastrointestinal tract. In contrast, false-negative 18F-FDG uptake was detected in 24 patients (58.5%), while false-positive 18F-FDG uptake was detected in 5 patients (12.2%). In the former case, 2/19 patients had both 18F-FDG-positive lesions and 18F-FDG-negative lesions in the gastrointestinal tract. In patients with 18F-FDG avidity, the SUVmax value of the involved gastrointestinal tract ranged from 2.6 to 17.4 (median: 4.7). For the 18F-FDG-negative (n = 22) and -positive (n = 19) groups, there were no differences in the male to female ratios (10/12 vs 4/15, P = 0.186), patient age (63.6 ± 2.4 years vs 60.1 ± 2.6 years, P = 0.323), presence of histopathological grade 1 vs 2 (20/2 and 17/2, P = 1.000), follicular dendritic cell pattern (duodenal/nodal: 13/5 vs 10/3, P = 1.000), mitotic rate (low/partly high, 14/1 vs 10/3, P = 0.311), clinical stage according to the Ann Arbor system (stages IE and IIE/other, 15/7 vs 15/4, P = 0.499), clinical stage according to the Lugano system (stages I and II-1/other, 14/8 vs 14/5, P = 0.489), soluble interleukin-2 receptor levels (495 ± 78 vs 402 ± 83, P = 0.884), LDH levels (188 ± 7 vs 183 ± 8, P = 0.749), hemoglobin levels (13.5 ± 0.3 vs 12.8 ± 0.4, P = 0.197), bone marrow involvement (positive/negative, 1/8 vs 1/10, P = 1.000), detectability by CT scanning (positive/negative, 1/16 vs 4/13, P = 0.335), and FLIPI risk (low risk/other, 16/6 vs 13/6, P = 0.763), respectively in each case. CONCLUSION: These findings indicate that it is not feasible to predict 18F-FDG-avidity. Therefore, 18F-FDG-PET scans represent a complementary modality for the detection of gastrointestinal involvements in follicular lymphoma patients, and surveillance of the entire gastrointestinal tract by endoscopic examinations is required.