Computer-aided diagnosis for optical diagnosis of diminutive colorectal polyps including sessile serrated lesions: a real-time comparison with screening endoscopists.

BACKGROUND : We aimed to compare the accuracy of the optical diagnosis of diminutive colorectal polyps, including sessile serrated lesions (SSLs), between a computer-aided diagnosis (CADx) system and endoscopists during real-time colonoscopy. METHODS : We developed the POLyp Artificial Recognition (POLAR) system, which was capable of performing real-time characterization of diminutive colorectal polyps. For pretraining, the Microsoft-COCO dataset with over 300 000 nonpolyp object images was used. For training, eight hospitals prospectively collected 2637 annotated images from 1339 polyps (i. e. publicly available online POLAR database). For clinical validation, POLAR was tested during colonoscopy in patients with a positive fecal immunochemical test (FIT), and compared with the performance of 20 endoscopists from eight hospitals. Endoscopists were blinded to the POLAR output. Primary outcome was the comparison of accuracy of the optical diagnosis of diminutive colorectal polyps between POLAR and endoscopists (neoplastic [adenomas and SSLs] versus non-neoplastic [hyperplastic polyps]). Histopathology served as the reference standard. RESULTS : During clinical validation, 423 diminutive polyps detected in 194 FIT-positive individuals were included for analysis (300 adenomas, 41 SSLs, 82 hyperplastic polyps). POLAR distinguished neoplastic from non-neoplastic lesions with 79 % accuracy, 89 % sensitivity, and 38 % specificity. The endoscopists achieved 83 % accuracy, 92 % sensitivity, and 44 % specificity. The optical diagnosis accuracy between POLAR and endoscopists was not significantly different (P = 0.10). The proportion of polyps in which POLAR was able to provide an optical diagnosis was 98 % (i. e. success rate). CONCLUSIONS : We developed a CADx system that differentiated neoplastic from non-neoplastic diminutive polyps during endoscopy, with an accuracy comparable to that of screening endoscopists and near-perfect success rate.

Bowel intussusception caused by a percutaneously placed endoscopic gastrojejunostomy catheter: A case report.

BACKGROUND: In adults, bowel intussusception is a rare diagnosis and is mostly due to an organic bowel disorder. In rare cases, this is a complication of a percutaneously placed endoscopic gastro (jejunostomy) catheter. CASE SUMMARY: We describe a case of a 73-year-old patient with a history of myocardial infarction, chronic idiopathic constipation and Parkinson's disease. For the admission of his Parkinson's medication, a percutaneous endoscopic gastrostomy with jejunal extension (PEG-J) was placed. The patient presented three times at the emergency department of the hospital with intermittent abdominal pain with nausea and vomiting. There were no distinctive abnormalities from the physical and laboratory examinations. An abdominal computed tomography scan showed a small bowel intussusception. By push endoscopy, a jejunal bezoar at the tip of the PEG-J catheter was found to be the cause of small bowel intussusception. The intussusception was resolved after removing the bezoar during push enteroscopy. CONCLUSION: Endoscopic treatment of bowel intussusception caused by PEG-J catheter bezoar.

Limited diagnostic value of microsatellite instability associated pathology features in colorectal cancer.

To determine the diagnostic test characteristics and inter-observer variation of pathology features for identifying high microsatellite instability (MSI-H) colorectal cancer (CRC). Six pathologists blindly evaluated 177 CRC for the presence of MSI-H associated pathology features. Inter-observer agreement was determined by using Kappa-statistics. In the first random 88/177 cases, mucinous carcinoma, tumor-infiltrating lymphocytes (TIL) and Crohns-like infiltrate (CLI) were the best discriminators between MSI-H and microsatellite stable CRC [OR 5.6 (95 % CI 1.7-19), 5.4 (1.8-17) and 3.5 (1.1-11), respectively], with high specificity (89-91 %). The sensitivities for MSI-H, however, were low (31-41 %). In addition, inter-observer agreement was moderate for TIL and CLI (κ 0.38 and 0.48, respectively), but very good for mucinous carcinoma (κ 0.86). Interpretation of overall histopathology as suggestive for MSI-H performed better than any individual feature; OR 15 (5.2-44), and area under the curve 0.79. However, inter-observer agreement was moderate (κ 0.53). In the second set, TIL and CLI were scored according to updated scoring systems. Although both remained the best individual discriminators, test characteristics and inter-observer agreement did not improve. MSI-H pathology features have moderate accuracy for identifying MSI-H CRC, and are identified with moderate inter-observer agreement. These findings highlight the limitations of clinical strategies, such as the revised Bethesda guidelines, which incorporate the MSI-H associated pathology features in their strategy to identify persons with lynch syndrome.

Comparing quality, safety, and costs of colonoscopies performed by nurse vs physician trainees.

BACKGROUND & AIMS: We evaluated the quality and safety of colonoscopies performed by nurse and physician endoscopy trainees as well as the cost differences. METHODS: We performed a study of 7 nurse and 8 physician (gastroenterology fellows) endoscopy trainees at 2 medical centers in the Netherlands from September 2008 through April 2012. At the beginning of the study, the subjects had no experience in endoscopy; they were trained in gastrointestinal endoscopy according to the regulations of the Dutch Society of Gastroenterology, performing a minimum of 100 colonoscopies. Each trainee then performed 135 consecutive colonoscopies (866 total by nurse trainees and 1080 by physician trainees) under supervision of a gastroenterologist; the colonoscopies were evaluated for quality and safety. We performed statistical analyses of data, assessing multilevel and cost minimization. The mean age of the patients was 57 years, and about half were women in each group. RESULTS: The endoscopic quality and safety were comparable between nurse and physician trainees. Overall rates of cecal intubation were 95% for nurses and 93% for physicians (P = .38), including procedures that required assistance from a supervisor; mean withdrawal times were 10.4 and 9.8 minutes, respectively (P = .44). Each group detected 27% of adenomas and had a 0.5% rate of complication. In both groups, the rates of unassisted cecal intubation gradually increased with the number of colonoscopies performed, from 70% for nurses and 74% for physicians at the beginning to 89% and 86%, respectively, at the end of the assessment period. Using a strategy in which 1 gastroenterologist supervises 3 nurses, the personnel costs decreased from $64.65 to $54.58. CONCLUSIONS: In a supervised setting, nurse endoscopists perform colonoscopies according to quality and safety standards that are comparable with those of physician endoscopist and can substantially reduce costs.

Nurse-led follow-up at home vs. conventional medical outpatient clinic follow-up in patients with incurable upper gastrointestinal cancer: a randomized study.

CONTEXT: Upper gastrointestinal cancer is associated with a poor prognosis. The multidimensional problems of incurable patients require close monitoring and frequent support, which cannot sufficiently be provided during conventional one to two month follow-up visits to the outpatient clinic. OBJECTIVES: To compare nurse-led follow-up at home with conventional medical follow-up in the outpatient clinic for patients with incurable primary or recurrent esophageal, pancreatic, or hepatobiliary cancer. METHODS: Patients were randomized to nurse-led follow-up at home or conventional medical follow-up in the outpatient clinic. Outcome parameters were quality of life (QoL), patient satisfaction, and health care consumption, measured by different questionnaires at one and a half and four months after randomization. As well, cost analyses were done for both follow-up strategies in the first four months. RESULTS: In total, 138 patients were randomized, of which 66 (48%) were evaluable. At baseline, both groups were similar with respect to clinical and sociodemographic characteristics and health-related QoL. Patients in the nurse-led follow-up group were significantly more satisfied with the visits, whereas QoL and health care consumption within the first four months were comparable between the two groups. Nurse-led follow-up was less expensive than conventional medical follow-up. However, the total costs for the first four months of follow-up in this study were higher in the nurse-led follow-up group because of a higher frequency of visits. CONCLUSION: The results suggest that conventional medical follow-up is interchangeable with nurse-led follow-up. A cost utility study is necessary to determine the preferred frequency and duration of the home visits.

Inter-observer variation in the histological diagnosis of polyps in colorectal cancer screening.

AIM: To determine the inter-observer variation in the histological diagnosis of colorectal polyps. METHODS AND RESULTS: Four hundred and forty polyps were randomly selected from a colorectal cancer screening programme. Polyps were first evaluated by a general (324 polyps) or expert (116 polyps) pathologist, and subsequently re-evaluated by an expert pathologist. Conditional agreement was reported, and inter-observer agreement was determined using kappa statistics. In 421/440 polyps (96%), agreement for their non-adenomatous or adenomatous nature was obtained, corresponding to a very good kappa value of 0.88. For differentiation of adenomas as non-advanced and advanced, consensus was obtained in 266/322 adenomas (83%), with a moderate kappa value of 0.58. For the non-adenomatous or adenomatous nature, both general and expert pathologists, and expert pathologists between each other, showed very good agreement {kappa values of 0.89 [95% confidence interval (CI) 0.83-0.95] and 0.86 (95% CI 0.73-0.98), respectively}. For categorization of adenomas as non-advanced and advanced, moderate agreement was found between general and expert pathologists, and between expert pathologists [kappa values of 0.56 (95% CI 0.44-0.67) and 0.64 (95% CI 0.43-0.85), respectively]. CONCLUSIONS: General and expert pathologists demonstrate very good inter-observer agreement for differentiating non-adenomas from adenomas, but only moderate agreement for non-advanced and advanced adenomas. The considerable variation in differentiating non-advanced and advanced adenomas suggests that more objective criteria are required for risk stratification in screening and surveillance guidelines.

A back-to-back comparison of white light video endoscopy with autofluorescence endoscopy for adenoma detection in high-risk subjects.

OBJECTIVE: To compare the sensitivity of autofluorescence endoscopy (AFE) and white light video endoscopy (WLE) for the detection of colorectal adenomas in high-risk patients belonging to Lynch syndrome (LS) or familial colorectal cancer (CRC) families. METHODS: This was a prospective single-centre study carried out in a tertiary referral centre. The subjects were 75 asymptomatic patients originating from LS or familial CRC families. Patients were examined with either WLE followed by AFE or AFE followed by WLE. Back-to-back colonoscopy was performed by two blinded endoscopists. All lesions were removed during the second endoscopic procedure. Lesions missed during the second procedure were identified and removed on third pass. The sensitivity calculations for colorectal adenomas were based on histology results. The main outcome measures were the difference in sensitivity between WLE and AFE for the detection of adenomas in patients with LS or familial CRC. RESULTS: At least one adenoma was detected in 41 (55%) patients. WLE identified adenomas in 28/41 patients and AFE in 37/41 patients, corresponding to a 32% increase. In total 95 adenomas were detected, 65 by WLE and 87 by AFE, resulting in a significantly higher sensitivity of AFE compared with WLE (92% vs 68%; p=0.001). The additionally detected adenomas with AFE were significantly smaller than the adenomas detected by WLE (mean 3.0 mm vs 4.9 mm, p<0.01). CONCLUSIONS: AFE improves the detection of colorectal adenomas in patients with LS or familial CRC. The results of this study suggest that AFE may be preferable for surveillance of these high-risk patients.