Accuracy of polyp localization at colonoscopy

Costs and benefits of a formal quality framework for colonoscopy: Economic evaluation

Background and study aims Reduction of colorectal cancer morbidity and mortality is one of the primary objectives of colonoscopy. Post-colonoscopy colorectal cancers (PCCRCs) are critical outcome parameters. Analysis of PCCRC rates can validate quality assurance measures in colonoscopy. We assessed the effectiveness of implementing a gastroenterologist-led quality framework that monitors key procedure quality indicators (i.e., bowel preparation quality, adenoma detection rates, or patient satisfaction) by comparing the PCCRC rate before and after implementation. Patients and methods Individuals who had a colonoscopy between 2010 and 2017 at a single tertiary center in Queensland, Australia, were included and divided into two groups: baseline (2010-2014) and redesign phase (2015-2017). Data linkage of the state-wide cancer registry and hospital records enabled identification of subjects who developed colorectal cancers within 5 years of a negative colonoscopy. Costs associated with quality improvement were assessed for effectiveness. Results A total of 19,383 individuals had a colonoscopy during the study period. Seventeen PCCRCs were detected. The PCCRC rate was 0.376 per 1,000 person-years and the average 5-year PCCRC risk ranged from 0.165% to 0.051%. The rate of PCCRCs was higher at the beginning (0.166%; 95% confidence interval [CI] 0.15%-0.17%) compared with the later period with full implementation of quality control measures (0.027%; 95% CI 0.023%-0.03%). The quality process determined an incremental cost-effectiveness ratio of -$5,670.53 per PCCRC avoided. Conclusions This large cohort study demonstrated that a formal gastroenterologist-led quality assurance framework embedded into the routine operations of a clinical department not only reduces interval cancers but is also cost-effective regarding life years gained and quality-adjusted life years.

Postcolonoscopy colorectal cancer: Prevalence, categorization and root-cause analysis based on the World Endoscopic Organization system

AIMS

The World Endoscopy Organization (WEO) recommends that endoscopy units implement a process to identify postcolonoscopy colorectal cancer (PCCRC). The aims of this study were to assess the 3-year PCCRC rate and to perform root-cause analyses and categorization in accordance with the WEO recommendations.

PATIENTS AND METHODS

Cases of colorectal cancers (CRCs) in a tertiary care center were retrospectively included from January 2018 to December 2019. The 3-year and 4-year PCCRC rates were calculated. A root-cause analysis and categorization of PCCRCs (interval and type A, B, C noninterval PCCRCs) were performed. The level of agreement between two expert endoscopists was assessed.

RESULTS

A total of 530 cases of CRC were included. A total of 33 were deemed PCCRCs (age 75.8±9.5 years; 51.5% women). The 3-year and 4-year PCCRC rates were 3.4% and 4.7%, respectively. The level of agreement between the two endoscopists was acceptable either for the root-cause analysis (k=0.958) or for the categorization (k=0.76). The most plausible explanations of the PCCRCs were 8 "likely new PCCRCs", 1 (4%) "detected, not resected", 3 (12%) "detected, incomplete resection", 8 (32%) "missed lesion, inadequate examination", and 13 (52%) "missed lesion, adequate examination". Most PCCRCs were deemed noninterval Type C PCCRCs (N=17, 51.5%).

CONCLUSION

WEO recommendations for root-cause analysis and categorization are useful to detect areas for improvement. Most PCCRCs were avoidable and were likely due to missed lesions during an otherwise adequate examination.

Metachronous Colorectal Adenomas Occur Close to the Index Lesion

GOALS

The aim of this study is to assess the spatial relationship between index and metachronous colorectal adenoma location.

BACKGROUND

After the complete excision of a human sporadic colorectal adenoma, patients are at elevated risk of developing a further metachronous adenoma. Data regarding the occurrence site of a metachronous colorectal adenoma relative to the index adenoma are scarce.

STUDY

Prospectively maintained databases were interrogated to identify all colonoscopies and adenoma excisions performed over a 10-year period at a single university teaching hospital. Data for the colonic segments at which adenoma removal were reported at index and all subsequent colonoscopies were extracted and 2 allied data sets merged.

RESULTS

A total of 15,121 colonoscopies and 4759 polyp events were recorded. Four hundred fifty-two patients [296 male, 156 female, median (range) age 75 (32 to 100) y] developed at least 1 metachronous adenoma at follow-up colonoscopy. When single index events only are considered (ie, synchronous adenoma cases excluded), over 61% of metachronous adenomas were recorded in the same or an adjacent colonic segment. When the full span of the colon is considered, metachronous adenomas were more likely to occur in a section of the colon proximal to that of the index adenoma (41%±5%) than the same (39%±5%) or distal segment (20%±5%; P =0.006; 1-way χ 2 test).

CONCLUSIONS

A metachronous human sporadic colorectal adenoma is more likely to be found in the same colonic segment to that of the index adenoma or 1 immediately adjacent. These data suggest a shared origin of metachronous adenoma with preceding lesions, supporting the existence of precancerous fields.

Artificial Intelligence Applied to Colonoscopy: Is It Time to Take a Step Forward?

Growing evidence indicates that artificial intelligence (AI) applied to medicine is here to stay. In gastroenterology, AI computer vision applications have been stated as a research priority. The two main AI system categories are computer-aided polyp detection (CADe) and computer-assisted diagnosis (CADx). However, other fields of expansion are those related to colonoscopy quality, such as methods to objectively assess colon cleansing during the colonoscopy, as well as devices to automatically predict and improve bowel cleansing before the examination, predict deep submucosal invasion, obtain a reliable measurement of colorectal polyps and accurately locate colorectal lesions in the colon. Although growing evidence indicates that AI systems could improve some of these quality metrics, there are concerns regarding cost-effectiveness, and large and multicentric randomized studies with strong outcomes, such as post-colonoscopy colorectal cancer incidence and mortality, are lacking. The integration of all these tasks into one quality-improvement device could facilitate the incorporation of AI systems in clinical practice. In this manuscript, the current status of the role of AI in colonoscopy is reviewed, as well as its current applications, drawbacks and areas for improvement.

Computer-aided classification of colorectal segments during colonoscopy: a deep learning approach based on images of a magnetic endoscopic positioning device

OBJECTIVE

Assessment of the anatomical colorectal segment of polyps during colonoscopy is important for treatment and follow-up strategies, but is largely operator dependent. This feasibility study aimed to assess whether, using images of a magnetic endoscope imaging (MEI) positioning device, a deep learning approach can be useful to objectively divide the colorectum into anatomical segments.

METHODS

Models based on the VGG-16 based convolutional neural network architecture were developed to classify the colorectum into anatomical segments. These models were pre-trained on ImageNet data and further trained using prospectively collected data of the POLAR study in which endoscopists were using MEI (3930 still images and 90,151 video frames). Five-fold cross validation with multiple runs was used to evaluate the overall diagnostic accuracies of the models for colorectal segment classification (divided into a 5-class and 2-class colorectal segment division). The colorectal segment assignment by endoscopists was used as the reference standard.

RESULTS

For the 5-class colorectal segment division, the best performing model correctly classified the colorectal segment in 753 of the 1196 polyps, corresponding to an overall accuracy of 63%, sensitivity of 63%, specificity of 89% and kappa of 0.47. For the 2-class colorectal segment division, 1112 of the 1196 polyps were correctly classified, corresponding to an accuracy of 93%, sensitivity of 93%, specificity of 90% and kappa of 0.82.

CONCLUSION

The diagnostic performance of a deep learning approach for colorectal segment classification based on images of a MEI device is yet suboptimal (clinicaltrials.gov: NCT03822390).

Utilization and reproducibility of World Endoscopy Organization post-colonoscopy colorectal cancer algorithms: retrospective analysis

BACKGROUND

Colorectal cancer (CRC) diagnosed following a cancer-negative colonoscopy is termed post-colonoscopy CRC (PCCRC). In addition to calculating PCCRC rates, the World Endoscopy Organization (WEO) recommends review of individual PCCRC cases, including categorization into interval/non-interval PCCRCs, and root cause analysis to determine the most plausible explanation. We aimed to test the usability, reproducibility, and outcomes of the WEO algorithms.

METHODS

All CRC cases diagnosed from January 2015 to December 2016 in a single organization were cross referenced with local endoscopy and pathology databases to identify cases of PCCRC. We assessed: 1) WEO most plausible explanation for PCCRC; and 2) WEO PCCRC interval/non-interval subtype categorization. Interobserver agreement was measured using Cohen's kappa (κ). Cases with interobserver variation underwent panel discussion to reach consensus.

RESULTS

Among 527 patients with CRC, 48 PCCRCs were identified. A consistent most plausible explanation was found in 97 % of cases, showing almost perfect agreement (κ = 0.94). Most PCCRCs (66 %) were attributed to "possible missed lesion, prior examination adequate." Interval/non-interval categorization was consistent in 77 %, showing substantial agreement (κ = 0.67). Following panel discussion, consensus was reached in all cases. Overall, 15 % were categorized as interval and 85 % as non-interval PCCRCs (12 % type A, 31 % type B, and 42 % type C).

CONCLUSIONS

Review of PCCRC cases using WEO recommendations was performed accurately at a local level using readily available clinical information. The high number of non-interval type B PCCRCs suggests a significant proportion of PCCRCs could be avoided by better adherence to recommended surveillance intervals.

Endoscopic clips allow for accurate pre-operative localisation of colorectal cancer

BACKGROUND

Colorectal cancer is a major cause of morbidity and mortality worldwide. Optimal management of this disease relies upon accurate pre-operative localisation to allow multidisciplinary discussion and treatment planning. Current pre-operative localisation methods consist of colonoscopy and computed tomography (CT), which are only 79%-85% accurate. To minimise this error, colonoscopy tattooing is a routine practice to facilitate operative localisation. The aim of this study is to investigate if endoscopic radiopaque clips can more accurately localise the lesions pre-operatively.

METHODS

A retrospective case-control study was conducted of patients diagnosed with colorectal cancer at a tertiary hospital between 2017 and 2019. Visualisation rates and accurate localisation rates were compared between patients receiving radiopaque clips and those who had colonoscopy alone. All patients received a tattoo distal to the tumour and a staging CT. Data on patient demographics, tumour demographics, post-procedure complications and changes to surgical management were collected.

RESULTS

Of 285 patients, 245 had tumour localisation with colonoscopy alone and 40 had additional clip localisation. Groups had comparable patient demographics. For patients receiving clips and follow-up CTs within 14 days, 92% of lesions were visualised and 100% of these lesions were accurately localised. In contrast, colonoscopy only accurately localised 77% of lesions (p < 0.01). This resulted in 1.2% of patients requiring an altered operation due to incorrect localisation. No clip-related complications were reported.

CONCLUSION

Radiopaque clips are a highly accurate and cost-effective method for localising colorectal cancer with a pre-operative accuracy rate over 92%.