A deep learning system for detection of early Barrett's neoplasia: a model development and validation study.

BACKGROUND: Computer-aided detection (CADe) systems could assist endoscopists in detecting early neoplasia in Barrett's oesophagus, which could be difficult to detect in endoscopic images. The aim of this study was to develop, test, and benchmark a CADe system for early neoplasia in Barrett's oesophagus. METHODS: The CADe system was first pretrained with ImageNet followed by domain-specific pretraining with GastroNet. We trained the CADe system on a dataset of 14 046 images (2506 patients) of confirmed Barrett's oesophagus neoplasia and non-dysplastic Barrett's oesophagus from 15 centres. Neoplasia was delineated by 14 Barrett's oesophagus experts for all datasets. We tested the performance of the CADe system on two independent test sets. The all-comers test set comprised 327 (73 patients) non-dysplastic Barrett's oesophagus images, 82 (46 patients) neoplastic images, 180 (66 of the same patients) non-dysplastic Barrett's oesophagus videos, and 71 (45 of the same patients) neoplastic videos. The benchmarking test set comprised 100 (50 patients) neoplastic images, 300 (125 patients) non-dysplastic images, 47 (47 of the same patients) neoplastic videos, and 141 (82 of the same patients) non-dysplastic videos, and was enriched with subtle neoplasia cases. The benchmarking test set was evaluated by 112 endoscopists from six countries (first without CADe and, after 6 weeks, with CADe) and by 28 external international Barrett's oesophagus experts. The primary outcome was the sensitivity of Barrett's neoplasia detection by general endoscopists without CADe assistance versus with CADe assistance on the benchmarking test set. We compared sensitivity using a mixed-effects logistic regression model with conditional odds ratios (ORs; likelihood profile 95% CIs). FINDINGS: Sensitivity for neoplasia detection among endoscopists increased from 74% to 88% with CADe assistance (OR 2·04; 95% CI 1·73-2·42; p<0·0001 for images and from 67% to 79% [2·35; 1·90-2·94; p<0·0001] for video) without compromising specificity (from 89% to 90% [1·07; 0·96-1·19; p=0·20] for images and from 96% to 94% [0·94; 0·79-1·11; ] for video; p=0·46). In the all-comers test set, CADe detected neoplastic lesions in 95% (88-98) of images and 97% (90-99) of videos. In the benchmarking test set, the CADe system was superior to endoscopists in detecting neoplasia (90% vs 74% [OR 3·75; 95% CI 1·93-8·05; p=0·0002] for images and 91% vs 67% [11·68; 3·85-47·53; p<0·0001] for video) and non-inferior to Barrett's oesophagus experts (90% vs 87% [OR 1·74; 95% CI 0·83-3·65] for images and 91% vs 86% [2·94; 0·99-11·40] for video). INTERPRETATION: CADe outperformed endoscopists in detecting Barrett's oesophagus neoplasia and, when used as an assistive tool, it improved their detection rate. CADe detected virtually all neoplasia in a test set of consecutive cases. FUNDING: Olympus.

Long-term oncological outcomes of endoscopic full-thickness resection after previous incomplete resection of low-risk T1 CRC (LOCAL-study): study protocol of a national prospective cohort study.

BACKGROUND: T1 colorectal cancer (CRC) without histological high-risk factors for lymph node metastasis (LNM) can potentially be cured by endoscopic resection, which is associated with significantly lower morbidity, mortality and costs compared to radical surgery. An important prerequisite for endoscopic resection as definite treatment is the histological confirmation of tumour-free resection margins. Incomplete resection with involved (R1) or indeterminate (Rx) margins is considered a strong risk factor for residual disease and local recurrence. Therefore, international guidelines recommend additional surgery in case of R1/Rx resection, even in absence of high-risk factors for LNM. Endoscopic full-thickness resection (eFTR) is a relatively new technique that allows transmural resection of colorectal lesions. Local scar excision after prior R1/Rx resection of low-risk T1 CRC could offer an attractive minimal invasive strategy to achieve confirmation about radicality of the previous resection or a second attempt for radical resection of residual luminal cancer. However, oncologic safety has not been established and long-term data are lacking. Besides, surveillance varies widely and requires standardization. METHODS/DESIGN: In this nationwide, multicenter, prospective cohort study we aim to assess feasibility and oncological safety of completion eFTR following incomplete resection of low-risk T1 CRC. The primary endpoint is to assess the 2 and 5 year luminal local tumor recurrence rate. Secondary study endpoints are to assess feasibility, percentage of curative eFTR-resections, presence of scar tissue and/or complete scar excision at histopathology, safety of eFTR compared to surgery, 2 and 5 year nodal and/or distant tumor recurrence rate and 5-year disease-specific and overall-survival rate. DISCUSSION: Since the implementation of CRC screening programs, the diagnostic rate of T1 CRC is steadily increasing. A significant proportion is not recognized as cancer before endoscopic resection and is therefore resected through conventional techniques primarily reserved for benign polyps. As such, precise histological assessment is often hampered due to cauterization and fragmentation and frequently leads to treatment dilemmas. This first prospective trial will potentially demonstrate the effectiveness and oncological safety of completion eFTR for patients who have undergone a previous incomplete T1 CRC resection. Hereby, substantial surgical overtreatment may be avoided, leading to treatment optimization and organ preservation. Trial registration Nederlands Trial Register, NL 7879, 16 July 2019 ( https://trialregister.nl/trial/7879 ).

Endoscopic Resection Without Subsequent Ablation Therapy for Early Barrett's Neoplasia: Endoscopic Findings and Long-Term Mortality.

INTRODUCTION: After endoscopic resection (ER) of neoplasia in Barrett's esophagus (BE), it is recommended to ablate the remaining BE to minimize the risk for metachronous disease. However, we report long-term outcomes for a nationwide cohort of all patients who did not undergo ablation of the remaining BE after ER for early BE neoplasia, due to clinical reasons or performance status. METHODS: Endoscopic therapy for BE neoplasia in the Netherlands is centralized in 8 expert centers with specifically trained endoscopists and pathologists. Uniformity is ensured by a joint protocol and regular group meetings. We report all patients who underwent ER for a neoplastic lesion between 2008 and 2018, without further ablation therapy. Outcomes include progression during endoscopic FU and all-cause mortality. RESULTS: Ninety-four patients were included with mean age 74 (± 10) years. ER was performed for low-grade dysplasia (LGD) (10%), high-grade dysplasia (HGD) (25%), or low-risk esophageal adenocarcinoma (EAC) (65%). No additional ablation was performed for several reasons; in 73 patients (78%), the main argument was expected limited life expectancy. Median C2M5 BE persisted after ER, and during median 21 months (IQR 11-51) with 4 endoscopies per patient, no patient progressed to advanced cancer. Seventeen patients (18%) developed HGD/EAC: all were curatively treated endoscopically. In total, 29/73 patients (40%) with expected limited life expectancy died due to unrelated causes during FU, none of EAC. CONCLUSION: In selected patients, ER monotherapy with endoscopic surveillance of the residual BE is a valid alternative to eradication therapy with ablation.

Stent-in-stent technique for removal of embedded esophageal self-expanding metal stents.

OBJECTIVES: Partially covered self-expanding metal stents (SEMSs) are regularly used for malignant and occasionally for benign esophageal disorders. Safe removal of these stents can be challenging due to embedding of the uncovered stent ends. Our aim is to report the results of removal of embedded, partially covered SEMSs by induction of pressure necrosis using the stent-in-stent technique. METHODS: Consecutive patients referred to three endoscopy units in 2007-2009, treated by the stent-in-stent technique, were reviewed. The partially covered SEMSs were inserted for malignant (n=3) or benign (n=16) conditions and were left in situ for a median of 42 days (14-189). When SEMSs were found to be embedded, a fully covered self-expanding plastic stent (SEPS) or fully covered SEMS was placed inside the partially uncovered SEMS. Subsequent removal of both stents was planned after a period of 10-14 days. RESULTS: In total, 23 stent-in-stent procedures were performed in 19 patients (10 males). Placement of a fully covered stent (SEPS: n=9 and SEMS: n=14) was technically successful in all patients. In 21 of 23 (91%) procedures, both stents were successfully removed in one procedure after a median of 12 days (5-18). In two patients, a repeat stent-in-stent procedure was needed for persistent embedding of the partially uncovered SEMSs. One (5%) procedure was complicated by severe bleeding, which could be treated endoscopically. In seven (36%) patients, the initial disorder had resolved after stent removal and no further endoscopic interventions were needed. Two (10%) patients were treated with chemoradiation or surgery for esophageal cancer after stent removal. In 10 (53%) patients, a repeat endoscopic intervention was required during follow-up because of progressive dysphagia or a persisting leak or fistula. CONCLUSIONS: The stent-in-stent technique is safe and effective for the removal of partially covered SEMSs that are embedded in the esophageal wall.