Readmission rate and complications following biopsy of the ampulla of Vater-A retrospective data analysis.

Background and Aim: Obtaining endoscopic biopsies from the ampulla of Vater is important for the diagnosis of lesions that are suspicious for neoplasia. The clinical safety profile is not well defined in the literature. Our aim was to evaluate the procedure-related readmission rate and complications from ampullary biopsy in patients undergoing duodenoscopy and endoscopic retrograde cholangiopancreatography (ERCP). Methods: A retrospective data analysis was performed on adult patients at Austin Hospital who underwent ampullary biopsies between 1 January 2010 and 1 March 12022. Medical records were identified using pathology databases. The electronic health record was reviewed for baseline characteristics including demographics, date, indication for ampullary biopsy, procedure type (duodenoscopy or ERCP), and procedural associated interventions during ERCP. Readmissions to the Austin Emergency Department within 30 days following the biopsy were identified, and complications were noted. Results: A total of 506 records were reviewed and 246 episodes of ampullary biopsy met the inclusion criteria. The procedure-related readmission rate for all episodes was 6.1%, which included pain (3.3%), pancreatitis (2.0%), cholangitis (1.6%), and bleeding (0.8%). Ampullary biopsies with ERCP had a procedure-related readmission rate of 8.4%, whereas ampullary biopsies without ERCP had a rate of 2.2%. Increased readmissions and complications were associated with male sex (P = 0.01 and P = 0.05, respectively). There was no association between the number of biopsies taken and complications. Conclusion: Performing an ampullary biopsy without an associated ERCP carries a low rate of clinical complications and procedure-related readmissions. The combination of ERCP and ampullary biopsy increases the risk four-fold.

Endoscopy training in Australia during COVID-19: Efficacy and knowledge assessment of gastroenterology and general surgery trainees.

Background and Aim: During COVID-19, restrictions to elective endoscopy were introduced worldwide. A reduction in procedures may impact trainees' endoscopy learning. This study aims to assess Australian advanced gastroenterology and general surgery trainees' self-perceived efficacy and knowledge in endoscopy during the pandemic. Methods: All Australian gastroenterology and general surgery trainees in their last 2 years of accredited training were invited to participate through email (2020-2021 and 2021-2022 training cycles). The primary outcome was to assess trainees' self-efficacy and knowledge regarding gastrointestinal endoscopy. Secondary outcomes included subgroup analysis between gastroenterology and general surgery trainees. Self-perceived efficacy was assessed with Likert-scale questions on 20 endoscopy procedures and knowledge was assessed through 21 endoscopy-related multiple choice questions. Results: Eighty-one trainees responded to a self-efficacy questionnaire and 77 responded to the knowledge questionnaire. Over 90% of the trainees were confident or extremely confident in diagnostic endoscopy, but only half demonstrated similar efficacy for therapeutic endoscopy. The efficacy for basic endoscopy procedures was higher for gastroenterology trainees (64.0% vs 51.1%, P < 0.001). Last-year trainee achievement of conjoint committee requirements for upper gastrointestinal endoscopy was achieved in 95.8% of gastroenterology trainees versus 22.2% of surgical trainees (P < 0.001). The median score on the knowledge questionnaire was also higher for the gastroenterology subset (90.5% vs 71.4%, P < 0.001). Conclusion: During COVID-19, endoscopy trainees' self-efficacy in endoscopic diagnostic procedures was achieved for most trainees. The differences in self-perceived efficacy and knowledge between gastroenterology and surgical trainees may be reflective of the different opportunities for learning between the two groups.

High-flow oxygen via oxygenating mouthguard in short upper gastrointestinal endoscopy: A randomised controlled trial.

BACKGROUND: Anaesthetic care during upper gastrointestinal (GI) endoscopy has the unique challenge of maintaining ventilation and oxygenation via a shared upper airway. Supplemental oxygen is recommended by international society guidelines, however, the optimal route or rate of oxygen delivery is not known. Various oxygen delivery devices have been investigated to improve oxygenation during upper GI endoscopy, however, these are limited by commercial availability, costs and in some cases, the expertise required for insertion. Anecdotally at our centre, higher flows of supplemental oxygen can safely be delivered via an oxygenating mouthguard routinely used during upper GI endoscopic procedures. AIM: To assess the incidence of hypoxaemia (SpO2 < 90%) in patients undergoing upper GI endoscopy receiving supplemental oxygen using an oxygenating mouthguard at 20 L/min flow compared to standard nasal cannula (SNC) at 2 L/min flow. METHODS: A single centre, prospective, randomised clinical trial at two sites of an Australian tertiary hospital between October 2020 and September 2021 was conducted. Patients undergoing elective upper gastrointestinal endoscopy under deep sedation were randomised to receive supplemental oxygen via high-flow via oxygenating mouthguard (HFMG) at 20 L/min flow or SNC at 2 L/min flow. The primary outcome was the incidence of hypoxaemia of any duration measured by pulse oximetry. Intraprocedural-related, procedural-related, and sedation-related adverse events and patient-reported outcomes were also recorded. RESULTS: Three hundred patients were randomised. Eight patients were excluded after randomisation. 292 patients were included in the intention-to-treat analysis. The incidence of hypoxaemia was significantly reduced in those allocated HFMG. Six patients (4.4%) allocated to HFMG experienced an episode of hypoxaemia, compared to thirty-four (22.1%) patients allocated to SNC (P value < 0.001). No significant difference was observed in the rates of adverse events or patient-reported outcome measures. CONCLUSION: The use of HFMG offers a novel approach to reducing the incidence of hypoxaemia during short upper gastrointestinal endoscopic procedures in low-risk patients undergoing deep sedation.

Endoscopic ultrasound-guided gallbladder and bile duct drainage with lumen apposing metal stent: A large multicenter cohort (with videos).

BACKGROUND AND AIM: Cholecystectomy and endoscopic retrograde cholangiopancreatography are the gold standard for managing acute cholecystitis and malignant biliary obstruction, respectively. Recent advances in therapeutic endoscopic ultrasound (EUS) have provided alternatives for managing patients in whom these approaches fail, namely, EUS-guided gallbladder drainage (EUS-GB) and EUS-guided bile duct drainage (EUS-BD). We aimed to assess the technical and clinical success of these techniques in the largest multicenter cohort published to date. METHODS: A retrospective, multicenter, observational study involving 17 centers across Australia and New Zealand was conducted. All patients who had EUS-GB or EUS-BD performed in a participating center using a lumen apposing metal stent between 2016 and 2020 were included. Primary outcome was technical success, defined as intra-procedural successful drainage. Secondary outcomes included clinical success and 30-day mortality. RESULTS: One hundred and fifteen patients underwent EUS-GB (n = 49) or EUS-BD (n = 66). EUS-GB was technically successful in 47 (95.9%) while EUS-BD was successful in 60 (90.9%). All failed cases were due to maldeployment of the distal flange outside of the targeted lumen. Clinical success of EUS-GB was achieved in 39 (79.6%). No patients required subsequent cholecystectomy. Clinical success of EUS-BD was achieved in 52 (78.8 %). Thirty-day mortality was 14.3% for EUS-GB and 12.1% for EUS-BD. CONCLUSIONS: EUS-guided gallbladder drainage and EUS-BD are promising alternatives for managing nonsurgical candidates with cholecystitis and malignant biliary obstruction following failed endoscopic retrograde pancreatography. Both techniques delivered high technical success with acceptable clinical success. Further research is needed to investigate the gap between technical and clinical success.

Long-term clinical outcomes of lipiodol marking using standard gastroscopy for image-guided radiotherapy of upper gastrointestinal cancers.

BACKGROUND: Image-guided radiotherapy (IGRT) has significantly improved the precision in which radiotherapy is delivered in cancer treatment. Typically, IGRT uses bony landmarks and key anatomical structures to locate the tumor. Recent studies have demonstrated the feasibility of peri-tumor fiducials in enabling even more accurate delineation of target and normal tissue. The use of gold coils as fiducials in gastrointestinal tumors has been extensively studied. However, placement requires expertise and specialized endoscopic ultrasound equipment. This article reports the long-term outcomes of using a standard gastroscopy to inject liquid fiducials for the treatment of oesophageal and gastric tumors with IGRT. AIM: To assess the long-term outcomes of liquid fiducial-guided IGRT in a cohort of oesophageal and gastric cancer patients. METHODS: A retrospective cohort study of consecutive adults with Oesophagogastric cancers referred for liquid fiducial placement before definitive/neo-adjuvant or palliative IGRT between 2013 and 2021 at a tertiary hospital in Melbourne, Australia was conducted. Up to four liquid fiducials were inserted per patient, each injection consisting of 0.2-0.5mL of a 1:1 mixture of iodized oil (Lipiodol; Aspen Pharmacare) and n-butyl 2-cyanoacrylate (Histoacryl®; B. Braun). A 23-gauge injector (Cook Medical) was used for the injection. All procedures were performed by or under the supervision of a gastroenterologist. Liquid fiducial-based IGRT (LF-IGRT) consisted of computer-assisted direct matching of the fiducial region on cone-beam computerised tomography at the time of radiotherapy. Patients received standard-IGRT (S-IGRT) if fiducial visibility was insufficient, consisting of bone match as a surrogate for tumor position. Radiotherapy was delivered to 54Gy in 30 fractions for curative patients and up to 45Gy in 15 fractions for palliative treatments. RESULTS: 52 patients were referred for liquid fiducial placement within the study period. A total of 51 patients underwent liquid fiducial implantation. Of these a total of 31 patients received radiotherapy. Among these, the median age was 77.4 years with a range between 57.5 and 88.8, and 64.5% were male. Twenty-seven out of the 31 patients were able to have LF-IGRT while four had S-IGRT. There were no complications after endoscopic implantation of liquid fiducials in our cohort. The cohort overall survival (OS) post-radiotherapy was 19 mo (range 0 to 87 mo). Whilst the progression-free survival (PFS) post-radiotherapy was 13 mo (range 0 to 74 mo). For those treated with curative intent, the median OS was 22.0 mo (range 0 to 87 mo) with a PFS median of 14.0 mo (range 0 to 74 mo). Grade 3 complication rate post-radiotherapy was 29%. CONCLUSION: LF-IGRT is feasible in 87.1% of patients undergoing liquid fiducial placement through standard gastroscopy injection technique. Our cohort has an overall survival of 19 mo and PFS of 13 mo. Further studies are warranted to determine the long-term outcomes of liquid-fiducial based IGRT.

Diagnostic accuracy of narrow-band imaging endoscopy with targeted biopsies compared with standard endoscopy with random biopsies in patients with Barrett's esophagus: A systematic review and meta-analysis.

BACKGROUND AND AIM: Endoscopic surveillance for dysplasia in Barrett's esophagus (BE) with random biopsies is the primary diagnostic tool for monitoring clinical progression into esophageal adenocarcinoma. As an alternative, narrow-band imaging (NBI) endoscopy offers targeted biopsies that can improve dysplasia detection. This study aimed to evaluate NBI-guided targeted biopsies' diagnostic accuracy for detecting dysplasia in patients undergoing endoscopic BE surveillance compared with the widely used Seattle protocol. METHODS: Cochrane DTA Register, MEDLINE/PubMed, EMBASE, OpenGrey, and bibliographies of identified papers were searched until 2018. Two independent investigators resolved discrepancies by consensus, study selection, data extraction, and quality assessment. Data on sensitivity, specificity, and predictive values were pooled and analyzed using a random-effects model. RESULTS: Of 9528 identified articles, six studies comprising 493 participants were eligible for quantitative synthesis. NBI-targeted biopsy showed high diagnostic accuracy in detection of dysplasia in BE with a sensitivity of 76% (95% confidence interval [CI]: 0.61-0.91), specificity of 99% (95% CI: 0.99-1.00), positive predictive value of 97% (95% CI: 0.96-0.99), and negative predictive value of 84% (95% CI: 0.69-0.99) for detection of all grades of dysplasia. The receiver-operating characteristic curve for NBI model performance was 0.8550 for detecting all dysplasia. CONCLUSION: Narrow-band imaging-guided biopsy demonstrated high diagnostic accuracy and might constitute a valid substitute for random biopsies during endoscopic surveillance for dysplasia in BE.

Video consultations during the coronavirus disease 2019 pandemic are associated with high satisfaction for both doctors and patients.

BACKGROUND AND AIM: Telehealth has become the standard of care during the COVID-19 outbreak. This study aimed to assess doctor and patient satisfaction of endoscopy-related telehealth clinics with video consultations. METHODS: A prospective observational study of patients consecutively booked to attend two endoscopy-related telehealth clinics at an ambulatory tertiary care setting was conducted from July to October 2020. Data collected from our previously published study using phone consultations (data collected in April-May 2020) were used as a control arm. The primary outcome (satisfaction) was assessed through the six-question score (6Q_score) as per previous research. Secondary outcomes included failure-to-attend (FTA) rate and perceived necessity of physical examination/in-person follow-up appointment. RESULTS: There were 962 endoscopy clinic appointments between July and October, of which 157 were conducted through video. Data on 127 doctor questionnaires and 94 patient questionnaires were analyzed. The median age (years) of patients reviewed via video [57, interquartile range (IQR) 48-66] was lower than those reviewed via phone (65, IQR 55-74, P < 0.01). Patient average 6Q_score was higher with video compared to phone (85.1% vs 78.4%, P = 0.01), as was doctors' 6Q_score (97.5% vs 91.9%, P = 0.02). FTA rates remained similar between the two assessments (6.4% in April/May and 4.4% between July/October, P = 0.12). The requirement for in-person follow-up/physical examination was identified in two video consultations (1.6%). CONCLUSION: Video consultations during the COVID-19 outbreak demonstrated higher patient and doctor satisfaction compared to phone consultations. There was no significant difference in FTA rates and need for in-person follow-up consultations/physical examination between the telehealth two modalities.

Benefits of phone consultation for endoscopy-related clinics in the COVID-19 pandemic.

BACKGROUND AND AIM: During COVID-19 outbreak, restrictions to in-person consultations were introduced with a rise in telehealth. An indirect benefit of telehealth could be better attendance. This study aimed to assess "failure-to-attend" (FTA) rate and satisfaction for two endoscopy-related compulsory telehealth clinics during the COVID-19 outbreak. METHODS: Consecutive patients booked for endoscopy-related telehealth clinics at a tertiary hospital were prospectively assessed. In-person clinic control data were assessed retrospectively. Sample size was calculated to detect an anticipated increase in attendance of 8%. Secondary outcomes included FTA differences between clinics and evaluation of patients and doctors satisfaction. Satisfaction was assessed based on six Likert scale questions used in previous telehealth research and asked to both patients and doctors (6Q_score). This study was exempt from IRB review after institutional IRB review. RESULTS: There were 691 patients booked for appointments in our endoscopy clinics during the study periods (373 in 2020). FTA rates were lowered by half during the compulsory telehealth clinics (12.6% to 6.4%, P < 0.01). The patient 6Q_score was higher for the advanced endoscopy clinic (84.6% vs 73.8%, P < 0.01), while the doctor 6Q_score was similar between both advanced clinics and post endoscopy clinics (91.1% vs 92.5% respectively, P = 0.80). An in-person follow-up consultation was suggested for 3.5% of the appointments, while the necessity of physical examination was flagged in 5.1%. CONCLUSIONS: The use of phone consultations in endoscopy-related clinics during the COVID-19 outbreak has improved FTA rates while demonstrating high satisfaction rates. The need for in-person follow-up consultations and physical examination were low.

Outcomes of endoscopic ultrasound as a one-off pancreatic cancer screening tool for 122 high- and moderate-risk patients.

BACKGROUND AND AIM: Pancreatic cancer (PC) carries a poor prognosis and is often detected at later stages. Screening programs for moderate- and high-risk people are still under debate. We present the results from a prospective study on endoscopic ultrasound (EUS) as a one-off screening tool for pancreatic cancer screening. METHODS: Asymptomatic patients with moderate- or high-risk of PC were invited to participate. Moderate risk consisted of one first-degree and at least one second-degree relative with PC and no PC-associated genetic mutations. High risk consisted of >1 first-degree relatives with PC or PC-associated mutations (i.e. BRCA2, Lynch Syndrome, Familial Atypical Multiple Mole Melanoma Syndrome, STK11, or PALB2). All included patients had genetic counseling and a screening EUS done. Primary outcome was the detection of PC on EUS. Secondary outcomes assessed the evolution of psychological symptoms based on the Impact of Events Scale (IES) and Personal Consequences Questionnaire (PCQ) before and after the screening took place. RESULTS: A total of 122 patients had a screening EUS performed between 2013 and 2019; 60 were male, 55.8 years was the mean age, 78 were at high risk for PC, and 25 had PC-associated mutations. No pancreatic cancers were identified at the one-off EUS screening. Overall, patients' IES/PCQ scores did not change after screening and feedback of no malignancy, with the exception of females (less concerned about PC after screening EUS). CONCLUSIONS: EUS did not detect any PCs in either a moderate- or high-risk population as a one-off screening method. The EUS procedure and genetic counseling improved psychological symptoms for the female subset of this population.

Comparison of different virtual chromoendoscopy classification systems for the characterization of colorectal lesions.

BACKGROUND AND AIM: Commonly used classifications for colorectal lesions (CLs) include the Narrow Band Imaging (NBI) International Colorectal Endoscopic (NICE) and Japan NBI Expert Team (JNET) classifications. However, both lack a sessile serrated adenoma/polyp (SSA/P) category. This has been addressed by the modified Sano's (MS) and Workgroup serrAted polypS and Polyposis (WASP) classifications. This study aims to compare the accuracy of wNICE and wJNET (WASP added to both) with the stand-alone MS classification. METHODS: Patients undergoing colonoscopy at an Australian tertiary hospital who had at least one CL detected were prospectively enrolled. In the exploratory phase, CLs were characterized in real time with NBI and magnification using all classifications. In the validation phase, CLs were assessed with both NBI and Blue Laser Imaging (BLI) by four external endoscopists in Japan. The primary outcome was the comparison of wJNET and MS. Secondary outcomes included comparisons among all classifications and the calculation of interrater reliability. RESULTS: A total of 483 CLs were evaluated in real time in the exploratory phase, and four sets of 30 CL images (80 on NBI and 40 on BLI) were scored in the validation phase. For high-confidence diagnoses, MS accuracy was superior to wJNET in both the exploratory (86% vs 79%, P < 0.05) and validation (85% vs 69%, P < 0.05) phases. The interrater reliability was substantial for all classifications (κ = 0.74, 0.69, and 0.63 for wNICE, wJNET, and MS, respectively). CONCLUSIONS: MS classification achieved the highest accuracy in both the exploratory and validation phases. MS can differentiate serrated and adenomatous polyps as a stand-alone classification.

Polyp Detection Rate as a Surrogate for Adenoma and Sessile Serrated Adenoma/Polyp Detection Rates.

INTRODUCTION: Quality measures for colonoscopy such as adenoma detection rate (ADR) have been proposed to be surveilled for ensuring minimum standards. However, its direct measurement is time consuming and often neglected. Extrapolating ADR and other quality measures from polyp detection rate (PDR) can be a pragmatic alternative. OBJECTIVE: To determine quotients for estimating ADR and sessile serrated adenoma/polyp detection rate (SSA/P-DR) from PDR in an Australian cohort. METHODS: Consecutive adult patient colonoscopies during a 1-year period were retrospectively assessed in a single Australian tertiary endoscopy center. Adenoma detection quotient (ADQ) and SSA/P detection quotient (SSA/P-DQ) were defined as the division of ADR and SSA/P-DR by PDR, respectively. The primary outcome was the number of procedures to achieve a stable cumulative ADQ and SSA/P-DQ. Secondary outcomes included evaluation of ADQ and SSA/P-DQ in different subsets. RESULTS: In total, 2,657 colonoscopies were performed by 15 endoscopists in 2016. The ADR, SSA/P-DR, and PDR found were 32.2, 6.7, and 47.3%, respectively. The ADQ and SSA/P-DQ values found were 0.68 and 0.14, respectively. After approximately 500 procedures, both ADQ and SSA/P-DQ became stable. Interclass correlation coefficient (ICC) for the prediction of ADR from ADQ was excellent for all endoscopists that performed >177 procedures in that year (ICC 0.84). CONCLUSIONS: ADQ and SSA/P-DQ values were consistent when over 500 procedures were analyzed. ADQ had an excellent correlation with ADR when >177 procedures per endoscopist were evaluated.

Learning curve for mastery of colorectal endoscopic submucosal dissection: Perspectives from a large Japanese cohort.

BACKGROUND AND AIM: Endoscopic submucosal dissection (ESD) is a challenging procedure. A dissection speed of ≥9 cm2/h has been acknowledged as a mark for expertise, alongside a complication rate of ≤5% and en bloc resection rate of ≥90%. However, there is lack of objective information on whether the three measures correlate with each other. This study aims to evaluate the dissection speed, safety, and efficacy of colorectal ESDs performed by experts and trainees. METHODS: Consecutive patients undergoing colorectal ESD at a Japanese hospital (2006-2017) were included in a prospectively collected database. Information on patient demographics, proceduralist, and intra-/postprocedure data was retrieved. The primary outcome was the comparison in dissection speed. The secondary outcomes included differences in safety and efficacy. Log-linear regression models adjusted for confounders (e.g. R0 resection) were used to assess the differences in dissection speed. RESULTS: Five hundred ninety procedures (514 patients) performed by 26 endoscopists were analyzed. Experts performed a higher number of difficult lesions (e.g. F2 fibrosis) but achieved higher dissection speed (10.3 vs 6.7 cm2/h). The difference was statistically significant for both unadjusted and adjusted models (P < 0.0001). The en bloc resection rates were similar for both groups (experts = 95.6%; trainees = 94.7%, P = 0.61). Although nonexperts damaged more of the muscularis propria (18.6 vs 12.5%, P = 0.04), this did not translate into a significant difference in perforation (experts = 3.7%; trainees = 6.9%, P = 0.09) or delayed bleeding (experts = 2.9%; trainees = 4.4%, P = 0.34). The dissection speed steadily increased with expertise. CONCLUSION: Although dissection speed for colorectal ESD was significantly higher for experts, ESDs could be safely and efficaciously performed by ESD trainees.

Dynamic functional clot formation in patients undergoing endoscopic mucosal resection.

BACKGROUND AND AIM: Electric burns can induce fibrinolytic activity. Endoscopic mucosal resection (EMR) is a minimally invasive technique utilizing electrocautery for resection of dysplastic lesions from the gastrointestinal tract. EMR's main complication is clinically significant postendoscopic bleeding. Currently, no studies have investigated the impact of electrocautery during EMR on the coagulation cascade by viscoelastic hemostatic assay. METHODS: Thrombelastometry was performed and plasminogen levels were measured on blood samples taken prior to EMR, within an hour following the procedure and 2 days post-EMR. Data were natural log-transformed and analyzed using repeated-measure analysis of covariance (ANCOVA) accounting for age, sex, body mass index (BMI) and site of EMR. RESULTS: Plasminogen levels decreased post-EMR (P = 0.001) and then increased 2 days post-EMR (P < 0.018). FIBTEM A10 and Maximum Clot Firmness, and EXTEM maximum lysis decreased an hour following EMR (P < 0.05 for all). These three measurements then increased 2 days post-EMR (P < 0.01 for all). There were no significant differences in thrombelastometry or plasminogen measures according to sex, age, BMI, or site of EMR. One patient experienced clinically significant postendoscopic bleeding at one-week post-EMR, with substantially decreased FIBTEM A10 and Maximum Clot Firmness at 2 days post-EMR. CONCLUSIONS: Decreased post-EMR plasminogen corresponds with reduced clot firmness and enhanced lysis affecting clot quality, strength, and stability. While further investigation in a larger sample is required to confirm the overall risk of clinically significant postendoscopic bleeding and mechanisms for plasminogen activation; this study highlights the potential utility of thrombelastometry in assessing fibrinolytic activity during EMR.

Different factors are associated with conventional adenoma and serrated colorectal neoplasia.

Current data shows there are differences in factors associated with colorectal neoplasia based on geographical location and cultural settings. There are no studies focusing on the association between environmental factors and colorectal polyps in Australia. The aim of this study was to prospectively evaluate the association of various factors with different colorectal neoplasia histology. We utilized a simplified one-page questionnaire for patients undergoing colonoscopy for information on age; gender; comorbidities; family history of colorectal cancer; physical activity; smoking; diet; alcohol intake; and body mass index. Factors were then evaluated for association with the presence of: (1) neoplastic lesions; (2) conventional adenomas; (3) neoplastic serrated polyps; (4) any lesions (past and present); and (5) hyperplastic polyps. 291 procedures and 260 patients were included. Factors with a p-value < 0.2 in a univariate regression were included in an initial multivariable regression model. Backwards elimination was then performed, removing one predictor at a time until only significant predictors remained. In the final multivariable model, age≥65, male gender, type-2 diabetes mellitus, active smoking and family history of colorectal cancer were found to be statistically significant predictors for the presence of colorectal neoplasia. However, the significant predictors found for conventional adenomas (older age, male gender and smoking) were different from the significant predictors for neoplastic serrated polyps (type-2 diabetes mellitus and family history of colorectal cancer). Older age, male gender, type-2 diabetes mellitus, and smoking were significantly associated with the presence of colorectal neoplasia. The factors associated with conventional adenomas differed from those associated with neoplastic serrated polyps.

Deep learning uncertainty and confidence calibration for the five-class polyp classification from colonoscopy.

There are two challenges associated with the interpretability of deep learning models in medical image analysis applications that need to be addressed: confidence calibration and classification uncertainty. Confidence calibration associates the classification probability with the likelihood that it is actually correct - hence, a sample that is classified with confidence X% has a chance of X% of being correctly classified. Classification uncertainty estimates the noise present in the classification process, where such noise estimate can be used to assess the reliability of a particular classification result. Both confidence calibration and classification uncertainty are considered to be helpful in the interpretation of a classification result produced by a deep learning model, but it is unclear how much they affect classification accuracy and calibration, and how they interact. In this paper, we study the roles of confidence calibration (via post-process temperature scaling) and classification uncertainty (computed either from classification entropy or the predicted variance produced by Bayesian methods) in deep learning models. Results suggest that calibration and uncertainty improve classification interpretation and accuracy. This motivates us to propose a new Bayesian deep learning method that relies both on calibration and uncertainty to improve classification accuracy and model interpretability. Experiments are conducted on a recently proposed five-class polyp classification problem, using a data set containing 940 high-quality images of colorectal polyps, and results indicate that our proposed method holds the state-of-the-art results in terms of confidence calibration and classification accuracy.

Computer-aided diagnosis for characterization of colorectal lesions: comprehensive software that includes differentiation of serrated lesions.

BACKGROUND AND AIMS: Endoscopy guidelines recommend adhering to policies such as resect and discard only if the optical biopsy is accurate. However, accuracy in predicting histology can vary greatly. Computer-aided diagnosis (CAD) for characterization of colorectal lesions may help with this issue. In this study, CAD software developed at the University of Adelaide (Australia) that includes serrated polyp differentiation was validated with Japanese images on narrow-band imaging (NBI) and blue-laser imaging (BLI). METHODS: CAD software developed using machine learning and densely connected convolutional neural networks was modeled with NBI colorectal lesion images (Olympus 190 series - Australia) and validated for NBI (Olympus 290 series) and BLI (Fujifilm 700 series) with Japanese datasets. All images were correlated with histology according to the modified Sano classification. The CAD software was trained with Australian NBI images and tested with separate sets of images from Australia (NBI) and Japan (NBI and BLI). RESULTS: An Australian dataset of 1235 polyp images was used as training, testing, and internal validation sets. A Japanese dataset of 20 polyp images on NBI and 49 polyp images on BLI was used as external validation sets. The CAD software had a mean area under the curve (AUC) of 94.3% for the internal set and 84.5% and 90.3% for the external sets (NBI and BLI, respectively). CONCLUSIONS: The CAD achieved AUCs comparable with experts and similar results with NBI and BLI. Accurate CAD prediction was achievable, even when the predicted endoscopy imaging technology was not part of the training set.

Narrow-band imaging for scar (NBI-SCAR) classification: from conception to multicenter validation.

BACKGROUND AND AIMS: Surveillance post-endoscopic resection (ER) currently warrants biopsy samples from the resection site scar in most cases, although clinical practice is variable. A classification with standard criteria for scars has not yet been established. We aimed to create and validate a novel classification for post-ER scars by using specific criteria based on advanced imaging. METHODS: Key endoscopic features for scars with and without recurrence were (1) dark brown color, elongated/branched pit pattern, and dense capillary pattern and (2) whitish, pale appearance, round/slightly large pits, and irregular sparse vessels. Scars were first assessed with high-definition white-light endoscopy (HD-WLE) followed by interrogation with narrow-band imaging (NBI). Scars with at least 2 concordant characteristics were diagnosed with "high confidence" for NBI for scar (NBI-SCAR) classification. The final endoscopic predictions were correlated with histopathology. The primary outcome was the difference in sensitivity between NBI-SCAR and HD-WLE predictions. Secondary outcomes included the validation of our findings in 6 different endoscopy settings (Australia, United States, Japan, Brazil, Singapore, and Malaysia). The validation took place in 2 sessions separated by 2 to 3 weeks, each with 10 one-minute videos of post-ER scars on underwater NBI with dual focus. Inter-rater and intrarater reliability were calculated with Fleiss' free-marginal kappa and Bennett et al. S score, respectively. RESULTS: One hundred scars from 82 patients were included. Ninety-five scars were accurately predicted with high confidence by NBI-SCAR in the exploratory phase. NBI-SCAR sensitivity was significantly higher compared with HD-WLE (100% vs 73.7%, P < .05). In the validation phase, similar results were found for endoscopists who routinely perform colonoscopies and use NBI (sensitivity of 96.4%). The inter-rater and intrarater reliability throughout all centers were, respectively, substantial (κ = .61) and moderate (average S = .52) for this subset. CONCLUSIONS: NBI-SCAR has a high sensitivity and negative predictive value for excluding recurrence for endoscopists experienced in colonoscopy and NBI. In this setting, this approach may help to accurately evaluate or resect scars and potentially mitigate the burden of unnecessary biopsy samples.

Microbiota profile is different for early and invasive colorectal cancer and is consistent throughout the colon.

BACKGROUND AND AIM: Microbiota have been associated with several diseases including colorectal cancer (CRC). This study aimed to evaluate the microbiota in early/invasive CRC utilizing stool and cytological brushes to determine differences in relative abundance (RA). METHODS: Colonoscopy patients referred for endoscopic submucosal dissection or previous to CRC surgery were prospectively enrolled. Stool was collected pre-bowel preparation; and brush samples were taken during colonoscopy (three regions). DNA extraction, 16S rRNA next generation sequencing, and biostatistics (qiime and stamp software packages) followed. Primary outcome was the difference in RA of the Fusobacterium genus between the groups. Secondary outcomes included analyses of other microbiota. RESULTS: Twenty-five patients were included, of which 14 had invasive cancer (≥ 1000 mm into the submucosa). The three major genera for invasive cancer were Bacterioides, Oribacterium, and Fusobacterium, whereas for early cancer were Oribacterium, Bacterioides, and Prevotella (decreasing order of RA). There was a significantly higher RA of Fusobacterium in the invasive cancer group (9.65% vs 0.95%, respectively, P < 0.001). The RA of all genera was similar throughout the colon. In addition to Fusobacterium, the genera Corynebacterium, Enterococcus, Neisseria, Porphyromonas, and Sclegelella showed statistically higher RA in the invasive cancer group. Conversely, the genera Oribacterium, Desulfovibrio, Clostridiales, and Lactobacillus showed lower RA in the invasive cancer group. CONCLUSIONS: The RA of Fusobacterium is higher with invasive CRC than in early CRC patients. In addition, five other bacteria genera were found to be increased, and four decreased in invasive CRC patients. The microbiota per patient was similar throughout the colon.

Effect of time of day and specialty on polyp detection rates in Australia.

BACKGROUND AND AIM: Adenoma detection rate (ADR) is an important quality metric in colonoscopy. However, there is conflicting evidence around factors that influence ADR. This study aims to investigate the effect of time of day and endoscopist background on ADR and sessile serrated adenoma/polyp detection rate (SSA/P-DR) for screening colonoscopies. METHODS: Consecutive patients undergoing colonoscopy in 2016 were retrospectively evaluated. Primary outcome was the effect of time of day and endoscopist specialty on screening ADR. Secondary outcomes included evaluation of the same factors on SSA/P-DR and other metrics and collinearity of ADR and SSA/P-DR. Linear regression models were used for association between ADR, time of day, and endoscopist background. Bowel preparation, endoscopist, session, patient age, and gender were adjusted for. Linear regression model was also used for comparing ADR and SSA/P-DR. Chi-square was used for difference of proportions. RESULTS: Two thousand six hundred fifty-seven colonoscopies, of which 558 were screening colonoscopies, were performed. The adjusted mean ADR (screening) was 36.8% in the morning compared with 30.5% in the afternoon (P < 0.0001) and was 36.8% for gastroenterologists compared with 30.4% for surgeons (P < 0.0001). For every 1-h delay in commencing the procedure, there was a reduction in mean ADR by 3.4%. Using a linear regression model, a statistically significant positive association was found between ADR and SSA/P-DR (P < 0.0001). CONCLUSIONS: Morning and afternoon sessions and gastroenterologists and surgeons achieved the minimum standards recommended for ADR. Afternoon lists and surgeons were associated with a lower ADR compared with morning and gastroenterologists, respectively. Additionally, SSA/P-DR showed collinearity with ADR.