Risk factors of missed early gastric cancer in endoscopic resected population: a retrospective, case-control study.

BACKGROUND: Missed early gastric cancer (MEGC) is prevalent during esophagogastroduodenoscopy (EGD), which is the first-line recommended strategy for detecting early gastric cancer (EGC). Hence, we explored the risk factors for MEGC and different types of MEGC, based on the endoscopic resected population. METHODS: This retrospective, case-control study was conducted at Nanjing Drum Tower Hospital (NJDTH). We included patients who were diagnosed with EGC during screening EGD, underwent endoscopic resection, and were confirmed by postoperative pathology at the NJDTH from January 2014 to December 2021, and classified them into different types according to the different root causes of misses. Univariable, multivariable, subgroup and propensity score analyses were used to explore the risk factors for MEGC and different types of MEGC. RESULTS: A total of 447 patients, comprising 345 with initially detected early gastric cancer (IDEGC) and 102 with MEGC, were included in this study. Larger size (≥ 1 cm) (OR 0.45, 95% CI 0.27-0.74, P = 0.002) and invasion depth of submucosa (OR 0.26, 95% CI 0.10-0.69, P = 0.007) were negatively associated with MEGC. Use of sedation (OR 0.32, 95% CI 0.20-0.52, P < 0.001) and longer observation time (OR 0.60, 95% CI 0.37-0.96, P = 0.034) exhibited protective effect on MEGC. CONCLUSIONS: Smaller and more superficial EGC lesions are more susceptible to misdiagnosis. The use of sedation and prolonged observation time during EGD could help reduce the occurrence of MEGC.

DDX3X interacts with SIRT7 to promote PD-L1 expression to facilitate PDAC progression.

Pancreatic ductal adenocarcinoma (PDAC) is recognized as the most aggressive and fatal malignancy. A previous study reported that PDAC patients who exhibit elevated levels of DDX3X have a poor prognosis and low overall survival rate. However, the underlying molecular mechanism remains unclear. This study aimed to investigate the specific roles of DDX3X in PDAC. Multiple bioinformatics analyses were used to evaluate DDX3X expression and its potential role in PDAC. In vitro and in vivo studies were performed to assess the effects of DDX3X on PDAC cell growth. Furthermore, Western blotting, quantitative PCR, immunohistochemistry, immunofluorescence, mass spectrometry, coimmunoprecipitation and multiplexed immunohistochemical staining were conducted to identify the specific regulatory mechanism in PDAC. The results verified that DDX3X expression is notably upregulated in the tumor tissue vs. normal tissue of PDAC patients. DDX3X knockdown markedly suppressed the proliferation, invasion and migration of PDAC cells in vitro and inhibited tumor growth in vivo. Conversely, overexpression of DDX3X induced the opposite effect. Further studies supported that the DDX3X protein can associate with sirtuin 7 (SIRT7) to stimulate PDAC carcinogenesis and progression. Furthermore, SIRT7 inhibition significantly impeded DDX3X-mediated tumor growth both ex vivo and in vivo. The results also revealed that programmed death ligand 1 (PD-L1) expression is positively correlated with DDX3X expression. These results reveal significant involvement of the DDX3X-SIRT7 axis in the initiation and advancement of PDAC and offer previously undiscovered therapeutic options for PDAC management.

Submucosal gland differentiation and implications in esophageal basaloid squamous cell carcinomas.

Esophageal basaloid squamous cell carcinoma (BSCC) is a heterogenous entity with multilineage differentiation. It lacks systematical analysis on submucosal gland differentiation (SGD) due to the histological diversity and low incidence of esophageal BSCC. This study aims to find the correlation of SGD and clinicopathological features. A total of 152 esophageal BSCCs were separated into three histological groups: pure, mixed, and borderline group. The clinicopathological features were compared between different groups. The prevalence of SGD was also compared between cases of different groups. A panel of antibodies were used to identify SGD. The pure group differed from the mixed and borderline groups in many aspects, lymph node metastasis (LNM), cancer embolus, perineural invasion, and advanced stage occurred less frequently in the pure group (P<0.01). The pure group had a better but statistically insignificant overall survival (P=0.097). The squamous cell carcinoma (SCC) component or focal squamous differentiation was present in metastatic lymph nodes in almost all mixed BSCCs (95.7%, 22/23) with LNM. The LNM rate of superficial (T1b) BSCCs (17.6%, 6/34) was comparable to that of superficial (T1b) SCCs (18.5%, 57/308). However, LNM exclusively occurred in superficial mixed (3/5) and borderline (3/10) BSCCs. The IHC results demonstrated a prevalence of SGD in pure group (77%, 43/56). SGD is considered to be a favorable factor, while the squamous differentiation or invasive SCC component is an adverse factor in esophageal BSCCs. Refinement of classification is a promising way to improve patient management.

Monoallelic deleterious MUTYH mutations generate colorectal cancer: A case report.

Here we reported a particular case of MUTYH-associated polyposis (MAP) that had only one rare heterozygous variant, but some particular clinical manifestations contributed to occur in this male patient by only one defective MUTYH allele were worth of further investigation. We reported a case of MAP. It is about a 33-year-old man with chief complaints of hematochezia who had multiple polyps that were found in his colon via colonoscopy. He followed his doctor's advice and performed a genetic analysis examination. Germline test was positive for a major heterozygous variant: chr1:45800165 on the MUTYH gene. MUTYH gene sequence analysis confirmed the following heterozygous variant: c.55CT (p.R19X) in exon 2 (ClinVar NM_001128425). Unfortunately, his mother and daughter have the ILK variant according to genetic analysis. However, this variant at the site was not detected in his father. Various types of polyps were found on repeated colonoscopy, which tended to become latent cancerous in the future. This case indicated that awareness of the risk of carcinogenesis of polyps in carriers of monoallelic variants might accordingly increase, and our understanding of the type of genetically related disease will be enhanced by us.

Comparison of Modified Cap-Assisted Endoscopic Mucosal Resection and Endoscopic Submucosal Dissection in Treating Intraluminal Gastric Gastrointestinal Stromal Tumor (≤20 mm).

INTRODUCTION: A modified cap-assisted endoscopic mucosal resection (mEMR-C), introduced in this study, was a novel variation of the standard EMR. We aimed to compare the outcomes of mEMR-C and endoscopic submucosal dissection (ESD) for the treatment of small (≤20 mm) intraluminal gastric gastrointestinal stromal tumors (gGISTs). METHODS: This retrospective study included 43 patients who underwent mEMR-C and 156 patients who received ESD at Nanjing Drum Tower Hospital. Baseline characteristics, adverse events, and clinical outcomes were compared between the 2 groups. Univariate and multivariable analyses were conducted to adjust for confounders. After propensity score matching using sex, year, location, and tumor size, outcomes were compared with 41 patients in each group. RESULTS: A total of 199 patients underwent endoscopic resection and the en bloc resection rate was 100%. The complete resection rate was comparable in both groups ( P = 1.000). Approximately 9.5% of all patients had a positive margin. There was no significant difference in positive margin for patients undergoing mEMR-C or ESD (9.3% vs 9.6%, P = 1.000). No difference in adverse events in both groups ( P = 0.724). The mEMR-C was associated with shorter operation time and lower cost than the ESD. Recurrence was reported in 2 patients at 1 and 5 years after ESD during a median follow-up of 62 months. No metastasis and disease-related death were identified in both groups. Propensity score matching analysis revealed similar results. DISCUSSION: The mEMR-C was found to be the preferable technique for small (≤20 mm) intraluminal gGISTs with shorter operation time and lower cost as compared with ESD.

Risk factors of perforation in gastric stromal tumors during endoscopic resection: a retrospective case-control study.

BACKGROUND AND AIMS: Perforation is a common complication during endoscopic resection (ER) of gastric gastrointestinal stromal tumors (gGISTs) associated with secondary infections, sepsis, hospitalization time and cost. However, the risk factors of perforation remain controversial. This study aimed to investigate the risk factors for perforation during ER of gGISTs. METHODS: This retrospective case-control study included consecutive patients with gGISTs who underwent ER between June 2009 and November 2021 at the Nanjing Drum Tower Hospital. Univariate and multivariate analyses were performed to investigate the risk factors for perforation. Sensitivity analyses with propensity scoring (PS) were performed to evaluate the stability of the independent effects. RESULTS: In total, 422 patients with gGISTs were included. The following factors were associated with perforation during ER: in the non-intraluminal growth patterns (all confounders adjusted odds ratio [aOR]: 5.39, 95% CI 2.99-9.72, P < 0.001), in the gastric fundus (aOR 2.25, 95% CI 1.40-3.60, P = 0.007), sized ≥ 2 cm (aOR 1.70, 95% CI 1.04-2.77, P = 0.035), in the lesser curvature (aOR 0.12, 95% CI 0.05-0.27, P < 0.001), and in the gastric cardia (aOR 0.13, 95% CI 0.04-0.50, P = 0.003). The PS analysis confirmed the stable independent effects of these identified risk factors. CONCLUSIONS: ERs of gGISTs in non-intraluminal growth patterns, in the gastric fundus, and with larger tumor size were independent risk factors for perforation. While tumors in the lesser curvature or gastric cardia were independent protective factor for perforation.

Sedation Is Associated with Higher Polyp and Adenoma Detection Rates during Colonoscopy: A Retrospective Cohort Study.

Background and Aims: Currently sedation is a common practice in colonoscopy to reduce pain of patients and improve the operator satisfaction, whereas its impact on examination quality, especially adenoma detection rate (ADR) is still controversial. Thus, we aimed to investigate the association of sedation with ADR. Methods: Consecutive patients receiving colonoscopy between January 2017 and January 2020 at the Nanjing Drum Tower Hospital, Nanjing, China, were collected. Univariate and multivariate logistic regression models were performed to investigate the association between sedation and ADR. Subgroup analysis and propensity score matching (PSM) analysis, as sensitivity analysis, were performed to validate the independent effect. Results: The ADR was significantly higher in cases with sedation (ADR: 36.9% vs. 29.1%, odds ratio [OR]: 1.42, 95% confidence interval [CI]: 1.31-1.55, P < 0.001). Multivariate analysis showed that the sedation was an independent factor associated with ADR (OR: 1.49, 95% CI: 1.35-1.65, P < 0.001). The effect was consistent in subgroup analyses (P > 0.05) and PSM analysis (ADR: 37.6% vs. 29.1%, OR: 1.47, 95% CI: 1.33-1.63, P < 0.001). Conclusion: Sedation was associated with a higher polyp and ADR s during colonoscopy, which can promote the quality of colonoscopy.

A real-time interpretable artificial intelligence model for the cholangioscopic diagnosis of malignant biliary stricture (with videos).

BACKGROUND AND AIMS: It is crucial to accurately determine malignant biliary strictures (MBSs) for early curative treatment. This study aimed to develop a real-time interpretable artificial intelligence (AI) system to predict MBSs under digital single-operator cholangioscopy (DSOC). METHODS: A novel interpretable AI system called MBSDeiT was developed consisting of 2 models to identify qualified images and then predict MBSs in real time. The overall efficiency of MBSDeiT was validated at the image level on internal, external, and prospective testing data sets and subgroup analyses, and at the video level on the prospective data sets; these findings were compared with those of the endoscopists. The association between AI predictions and endoscopic features was evaluated to increase the interpretability. RESULTS: MBSDeiT can first automatically select qualified DSOC images with an area under the curve (AUC) of .963 and .968 to .973 on the internal testing data set and the external testing data sets, and then identify MBSs with an AUC of .971 on the internal testing data set, an AUC of .978 to .999 on the external testing data sets, and an AUC of .976 on the prospective testing data set, respectively. MBSDeiT accurately identified 92.3% of MBSs in prospective testing videos. Subgroup analyses confirmed the stability and robustness of MBSDeiT. The AI system achieved superior performance to that of expert and novice endoscopists. The AI predictions were significantly associated with 4 endoscopic features (nodular mass, friability, raised intraductal lesion, and abnormal vessels; P < .05) under DSOC, which is consistent with the endoscopists' predictions. CONCLUSIONS: The study findings suggest that MBSDeiT could be a promising approach for the accurate diagnosis of MBSs under DSOC.

Comparison of Safety and Outcomes between Endoscopic and Surgical Resections of Intermediate-Risk Primary Gastric Gastrointestinal Stromal Tumors.

BACKGROUND AND AIMS: The objective of this study was to compare the safety and efficacy of endoscopic resection with surgical resection in the treatment of intermediate-risk gastric gastrointestinal stromal tumors (GISTs) and to further evaluate whether imatinib adjuvant treatment is necessary for resected intermediate-risk gastric GIST by ER. METHODS: We retrospectively studied 128 cases for intermediate-risk gastric GISTs that were distributed in endoscopic (n = 33) and surgical groups (n = 95) at our center between December 2009 to July 2020. We statistically compared the clinical features, pathological reports, perioperative data, and long-term follow-up outcomes. RESULTS: Compared with the surgery group, the endoscopy group was associated with smaller tumor size (2.4 ± 1.0 vs. 6.0 ± 1.7 cm, p < 0.001), shorter operating time (67.3 ± 36.5 vs. 145.9 ± 74.8 min, p < 0.001), fewer incidence of short-term postoperative complications (3% vs. 32.6%, p = 0.002). Shorter postoperative hospital days (4.5 ± 1.4 vs. 8.5 ± 2.4 days, p < 0.001), shorter gastric functional recovery time (p < 0.001), and a lower overall medical cost of hospitalization (p < 0.001) was detected in the endoscopy group. During the median 44.5 months follow-up period, there were no cases of recurrence, metastasis, and death in the endoscopy group. Among 128 patients, 68 accepted adjuvant therapy with imatinib after resection. It was observed that the OS of the adjuvant treatment group with imatinib was lower than that of the group without imatinib (p = 0.033). CONCLUSION: Endoscopic resection for intermediate-risk gastric GIST is a feasible and safe method, and there is no significant benefit for patients with intermediate-risk gastric GIST to accept imatinib adjuvant treatment after ER.

A Retrospective Cohort Study of Factors Influencing Lymph Node Metastasis in Patients With Early Gastric Papillary Adenocarcinoma.

INTRODUCTION: High risk of lymph node metastasis (LNM) in gastric papillary adenocarcinoma causes endoscopists to worry about the suitability of endoscopic resection for early gastric papillary adenocarcinoma (EPAC). We compared risk factors and attempted to establish a scoring system to stratify LNM risk in patients with EPAC. METHODS: A retrospective analysis was performed on 2,513 patients with early gastric carcinoma (EGC) who underwent radical resection in 4 tertiary hospitals in China. Univariate and multivariate analyses were performed to compare the invasiveness in EPAC and other types of EGC and to evaluate potential factors in predicting LNM risk in EPAC groups. RESULTS: Three hundred thirty-five patients with EPAC were enrolled in our study, of which 62 patients were found to have LNM. After comparing clinicopathological characteristics of EPAC with and without LNM, the following factors were included in the risk scoring system: 1 point each for lower stomach location and tumor size >2.0 cm, 3 points for lymphovascular invasion, and 4 points for submucosal invasion; the risk scoring system was validated in a small internal validation set with an area under the curve of 0.844. DISCUSSION: Our results suggested that EPAC was highly invasive compared with other EGCs, especially differentiated EGC types, and need to be treated more rigorously. This proposed risk scoring system could stratify LNM risk in patients with EPAC, and endoscopic resection may only be performed safely on the groups with a low LNM rate.

A deep learning-based model improves diagnosis of early gastric cancer under narrow band imaging endoscopy.

BACKGROUND: Diagnosis of early gastric cancer (EGC) under narrow band imaging endoscopy (NBI) is dependent on expertise and skills. We aimed to elucidate whether artificial intelligence (AI) could diagnose EGC under NBI and evaluate the diagnostic assistance of the AI system. METHODS: In this retrospective diagnostic study, 21,785 NBI images and 20 videos from five centers were divided into a training dataset (13,151 images, 810 patients), an internal validation dataset (7057 images, 283 patients), four external validation datasets (1577 images, 147 patients), and a video validation dataset (20 videos, 20 patients). All the images were labeled manually and used to train an AI system using You look only once v3 (YOLOv3). Next, the diagnostic performance of the AI system and endoscopists were compared and the diagnostic assistance of the AI system was assessed. The accuracy, sensitivity, specificity, and AUC were primary outcomes. RESULTS: The AI system diagnosed EGCs on validation datasets with AUCs of 0.888-0.951 and diagnosed all the EGCs (100.0%) in video dataset. The AI system achieved better diagnostic performance (accuracy, 93.2%, 95% CI, 90.0-94.9%) than senior (85.9%, 95% CI, 84.2-87.4%) and junior (79.5%, 95% CI, 77.8-81.0%) endoscopists. The AI system significantly enhanced the performance of endoscopists in senior (89.4%, 95% CI, 87.9-90.7%) and junior (84.9%, 95% CI, 83.4-86.3%) endoscopists. CONCLUSION: The NBI AI system outperformed the endoscopists and exerted potential assistant impact in EGC identification. Prospective validations are needed to evaluate the clinical reinforce of the system in real clinical practice.

A deep learning-based segmentation system for rapid onsite cytologic pathology evaluation of pancreatic masses: A retrospective, multicenter, diagnostic study.

BACKGROUND: We aimed to develop a deep learning-based segmentation system for rapid on-site cytopathology evaluation (ROSE) to improve the diagnostic efficiency of endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) biopsy. METHODS: A retrospective, multicenter, diagnostic study was conducted using 5345 cytopathological slide images from 194 patients who underwent EUS-FNA. These patients were from Nanjing Drum Tower Hospital (109 patients), Wuxi People's Hospital (30 patients), Wuxi Second People's Hospital (25 patients), and The Second Affiliated Hospital of Soochow University (30 patients). A deep convolutional neural network (DCNN) system was developed to segment cell clusters and identify cancer cell clusters with cytopathological slide images. Internal testing, external testing, subgroup analysis, and human-machine competition were used to evaluate the performance of the system. FINDINGS: The DCNN system segmented stained cells from the background in cytopathological slides with an F1-score of 0·929 and 0·899-0·938 in internal and external testing, respectively. For cancer identification, the DCNN system identified images containing cancer clusters with AUCs of 0·958 and 0·948-0·976 in internal and external testing, respectively. The generalizable and robust performance of the DCNN system was validated in sensitivity analysis (AUC > 0·900) and was superior to that of trained endoscopists and comparable to cytopathologists on our testing datasets. INTERPRETATION: The DCNN system is feasible and robust for identifying sample adequacy and pancreatic cancer cell clusters. Prospective studies are warranted to evaluate the clinical significance of the system. FUNDING: Jiangsu Natural Science Foundation; Nanjing Medical Science and Technology Development Funding; National Natural Science Foundation of China.

Diagnostic value of endoscopic ultrasound-guided fine needle aspiration with rapid on-site evaluation performed by endoscopists in solid pancreatic lesions: A prospective, randomized controlled trial.

BACKGROUND AND AIM: Endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) is the most established diagnostic method for pancreatic tissue. Rapid on-site evaluation by a trained endoscopist (self-ROSE) can improve the diagnostic accuracy. This research is aimed to analyze the application value of self-ROSE for EUS-FNA in solid pancreatic lesions. METHODS: A total of 194 consecutive patients with solid pancreatic lesions in Nanjing Drum Tower Hospital were randomized in a 1:1 ratio to EUS-FNA with or without self-ROSE in this single-center randomized controlled trial. Before initiating self-ROSE, the endoscopist underwent training for pancreatic cytologic sample adequacy assessment and cytopathological diagnosis of EUS-FNA in pathology department for 1 month. Some parts of the slides of EUS-FNA were air dried, stained on-site with BASO Liu's reagent, and on-site evaluated in self-ROSE group. Between the two groups, the diagnostic performance of EUS-FNA was analyzed, including sensitivity, specificity, positive predictive value, negative predictive value, and accuracy, with a comparison of the number of needle passes and the complication rates. RESULTS: The accuracy, sensitivity, specificity, positive predictive value, and negative predictive value were 94.8%, 94.4%, 100%, 100%, and 58.3% in the self-ROSE group, respectively, and 70.1%, 65.1%, 100%, 100%, and 32.6% in the non-self-ROSE group. The diagnostic accuracy (P < 0.001) and sensitivity (P < 0.001) were both significantly increased during EUS-FNA in the self-ROSE group compared to the non-self-ROSE group. The rate of cytologic sample adequacy was 100% in self-ROSE group and 80.4% in non-self-ROSE group. The number of passes were 3.38 ± 1.00 in self-ROSE group and 3.22 ± 0.89 in non-self-ROSE group (P = 0.228). No complications were found in both. There was acceptable consistency between endoscopist and pathologist in the cytopathological diagnosis (kappa = 0.666, P < 0.05) and in the sample adequacy rate (kappa = 1.000, P < 0.001). CONCLUSION: Our results demonstrated that self-ROSE is valuable for EUS-FNA in the diagnosis of solid pancreatic lesions and is an important choice to routinely increase the accuracy of EUS-FNA in centers without ROSE assessment.

Comparison of endoscopic full-thickness resection and cap-assisted endoscopic full-thickness resection in the treatment of small (≤1.5 cm) gastric GI stromal tumors.

BACKGROUND AND AIMS: With the increasing incidence of small GI stromal tumors (GISTs), endoscopic full-thickness resection (EFTR) and cap-assisted EFTR (EFTR-C) have been suggested as 2 effective resection methods. We aimed to compare the outcomes of EFTR and EFTR-C for the treatment of small (≤1.5 cm) gastric GISTs. METHODS: This retrospective study included 67 patients who underwent EFTR and 46 patients who underwent EFTR-C at Nanjing Drum Tower Hospital. Clinicopathologic features, adverse events (AEs), and outcomes were compared between the 2 groups. Univariate and multivariate linear and logistic regressions were used to analyze the effects of the procedure on the therapeutic outcomes of patients and adjusted for covariates in the multivariate analysis. RESULTS: The tumor size in the EFTR group tended to be larger (P = .005). The resection time in the EFTR-C group was shorter than that in the EFTR group (38.3 ± 20.7 minutes vs 15.0 ± 11.8 minutes, P < .001), which retained statistical significance with adjustment for the covariates (adjusted mean difference, 22.2; 95% confidence interval, 15.0-29.4; P < .001). The R0 resection rate of the EFTR group was 94.0% and of the EFTR-C group 97.8% (P = .355). The EFTR-C group was superior to the EFTR group in terms of perioperative therapeutic outcomes, AEs, and postoperative recovery. No recurrence occurred in the EFTR and EFTR-C groups. CONCLUSIONS: EFTR-C was found to be the preferable technique for small (≤1.5 cm) gastric GISTs with shorter operation times, lower AEs, faster postoperative recovery, and shorter hospitalization times as compared with EFTR.

Pancreatic and gut hormone responses to mixed meal test in post-chronic pancreatitis diabetes mellitus.

OBJECTIVE: More than one-third of chronic pancreatitis patients will eventually develop diabetes, recently classified as post-chronic pancreatitis diabetes mellitus (PPDM-C). This study was aimed to investigate the pancreatic and gut hormone responses to a mixed meal test in PPDM-C patients, compared with non-diabetic chronic pancreatitis (CP), and type 2 diabetes patients or healthy controls. DESIGN AND METHODS: Sixteen patients with PPDM-C, 12 with non-diabetic CP as well as 10 with type 2 diabetes and healthy controls were recruited. All participants underwent mixed meal tests, and blood samples were collected for measurements of blood glucose, C-peptide, insulin, glucagon, pancreatic polypeptide (PP), ghrelin, peptide YY, glucagon like peptide-1 (GLP-1) and gastric inhibitory peptide (GIP). Indices of insulin sensitivity and secretion were calculated. Repeated measures analysis of variance was performed. RESULTS: Participants with PPDM-C exhibited decreases in both fasting and postprandial responses of C-peptide (P < 0.001), insulin (P < 0.001), ghrelin (P < 0.001) and PYY (P = 0.006) compared to participants with type 2 diabetes and healthy controls. Patients with CP showed blunted glucagon, PP and incretin reactions, while the responses were increased in patients with PPDM-C compared to controls. The level of insulin sensitivity was higher for PPDM-C than type 2 diabetes (P < 0.01), however the indices for early/late-phase and overall insulin secretion (P < 0.01) were lower. CONCLUSIONS: Patients with PPDM-C are characterized by decreased C-peptide, insulin, ghrelin and PYY responses, and similar levels of glucagon, PP, GIP and GLP-1 compared to those with type 2 diabetes. The above findings, when confirmed in a larger population, may prove helpful to establish the diagnosis of PPDM-C, and should promote study on underlying pathophysiological mechanisms.

A Novel Deep Learning System for Diagnosing Early Esophageal Squamous Cell Carcinoma: A Multicenter Diagnostic Study.

INTRODUCTION: This study aims to construct a real-time deep convolutional neural networks (DCNNs) system to diagnose early esophageal squamous cell carcinoma (ESCC) with white light imaging endoscopy. METHODS: A total of 4,002 images from 1,078 patients were used to train and cross-validate the DCNN model for diagnosing early ESCC. The performance of the model was further tested with independent internal and external validation data sets containing 1,033 images from 243 patients. The performance of the model was then compared with endoscopists. The accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and Cohen kappa coefficient were measured to assess performance. RESULTS: The DCNN model had excellent performance in diagnosing early ESCC with a sensitivity of 0.979, a specificity of 0.886, a positive predictive value of 0.777, a negative predictive value of 0.991, and an area under curve of 0.954 in the internal validation data set. The model also depicted a tremendously generalized performance in 2 external data sets and exhibited superior performance compared with endoscopists. The performance of the endoscopists was markedly elevated after referring to the predictions of the DCNN model. An open-accessed website of the DCNN system was established to facilitate associated research. DISCUSSION: A real-time DCNN system, which was constructed to diagnose early ESCC, showed good performance in validation data sets. However, more prospective validation is needed to understand its true clinical significance in the real world.

A Novel Model Based on Deep Convolutional Neural Network Improves Diagnostic Accuracy of Intramucosal Gastric Cancer (With Video).

BACKGROUND AND AIMS: Prediction of intramucosal gastric cancer (GC) is a big challenge. It is not clear whether artificial intelligence could assist endoscopists in the diagnosis. METHODS: A deep convolutional neural networks (DCNN) model was developed via retrospectively collected 3407 endoscopic images from 666 gastric cancer patients from two Endoscopy Centers (training dataset). The DCNN model's performance was tested with 228 images from 62 independent patients (testing dataset). The endoscopists evaluated the image and video testing dataset with or without the DCNN model's assistance, respectively. Endoscopists' diagnostic performance was compared with or without the DCNN model's assistance and investigated the effects of assistance using correlations and linear regression analyses. RESULTS: The DCNN model discriminated intramucosal GC from advanced GC with an AUC of 0.942 (95% CI, 0.915-0.970), a sensitivity of 90.5% (95% CI, 84.1%-95.4%), and a specificity of 85.3% (95% CI, 77.1%-90.9%) in the testing dataset. The diagnostic performance of novice endoscopists was comparable to those of expert endoscopists with the DCNN model's assistance (accuracy: 84.6% vs. 85.5%, sensitivity: 85.7% vs. 87.4%, specificity: 83.3% vs. 83.0%). The mean pairwise kappa value of endoscopists was increased significantly with the DCNN model's assistance (0.430-0.629 vs. 0.660-0.861). The diagnostic duration reduced considerably with the assistance of the DCNN model from 4.35s to 3.01s. The correlation between the perseverance of effort and diagnostic accuracy of endoscopists was diminished using the DCNN model (r: 0.470 vs. 0.076). CONCLUSIONS: An AI-assisted system was established and found useful for novice endoscopists to achieve comparable diagnostic performance with experts.

Differential prognostic implications of gastric adenocarcinoma based on Lauren's classification: a Surveillance, Epidemiology, and End Results (SEER)-based cohort study.

BACKGROUND: Our study aims to analyze the association between Lauren's classification and gastric adenocarcinoma prognosis using comprehensive statistical analyses. METHODS: According to the selection criteria, patients were included from the Surveillance, Epidemiology, and End Results (SEER) database. Univariate and multivariate Cox regression, propensity score matching, and a multivariate competing risk model were used to investigate the association between Lauren's classification and prognosis. Subgroup analysis was used to investigate the role of confounding factors on the association between Lauren types and prognosis. RESULTS: After exclusion, a total of 20,218 patients from the SEER database were included, with 14,374 intestinal types and 5,844 diffuse types. The univariate Cox regression analysis revealed that the diffuse type had a poorer cancer-specific survival (CSS) rate [hazard ratio (HR), 1.44; 95% confidence interval (CI), 1.38-1.50]. After adjusting for confounding variables, the diffuse type also showed a higher risk of cancer-specific death (HR, 1.20; 95% CI, 1.15-1.20). Sensitivity analysis showed that after propensity score matching, the diffuse type had a poorer CSS rate (HR, 1.23; 95% CI, 1.10-1.36), and the competing risk model further validated these results [subdistribution HR (SHR), 1.32; 95% CI, 1.23-1.41]. Moreover, subgroup analysis demonstrated stable results in the subgroups, except for patients with T1 stage (HR, 1.06; 95% CI, 0.87-1.28) and a tumor size <2 cm (HR, 1.00; 95% CI, 0.83-1.21). CONCLUSIONS: Diffuse-type gastric adenocarcinoma had an overall poorer prognosis compared to the intestinal type. However, in patients with T1 stage and tumor size <2 cm, the diffuse type had a comparable survival rate with the intestinal type.