Colorectal cancer (CRC) is the third most common malignancy worldwide and the second leading cause of cancer related death. It can be prevented by endoscopic detection and complete resection of colorectal polyps. The JNET (Japanese NBI Expert Team) classification is clinically useful to predict the histology of large non-pedunculated colorectal polyps (LNPCPs) using narrow-band imaging at endoscopy. Japanese experts can reliably predict histology including the presence and depth of submucosal invasive cancer (SMI) using JNET with accuracy \>87%. On the other hand, the International Evaluation of Endoscopy classification-JNET (IEE-JNET) group demonstrated that ESGE and JGES endoscopists had sufficient accuracy for JNET 1 (93.0%) but insufficient accuracy for JNET 2A/B and 3 (respectively 62.1%, 55.1% and 85.1%). Reliably distinguishing between JNET 2A, 2B and 3 has a profound clinical relevance, since JNET 2A lesions can safely be resected using pEMR whereas JNET 2B lesions should be resected en-bloc (EMR or ESD) due to the increased risk of cancer and JNET 3 lesions are preferably treated with surgery due to the high risk of deeply invasive carcinoma and the necessity of lymph node resection. This study aims to validate a new simplified score, the Colorectal Regular-Irregular Score (CRIS) to fulfill the urgent need for a more effective and easier to use tool to predict LNPCPs histology. CRIS is a simplification of the JNET score which is mainly used by Japanese endoscopists or experts, recent evidence suggests its accuracy when used in everyday endoscopy in the Western world is insufficient. The investigators aim to compare JNET with CRIS for LNPCPs histology prediction amongst Western endoscopists using both original JNET interpretation and a clinically relevant approach. The study consists of three work packages (WPs): Work package one involves an expert online study where twelve expert endoscopists will evaluate 32 high-quality images of colorectal polyps using both JNET and CRIS classifications. Work package two involves an image/video-based online study where non-expert participants will be randomly assigned to rate images and videos using either JNET or CRIS, with performance re-evaluated after three months. Work package three involves a clinical study in a live endoscopy environment where non-expert endoscopists will participate in a randomized controlled trial assessing 10 colorectal polyps using either JNET or CRIS.
Study Type
INTERVENTIONAL
Allocation
RANDOMIZED
Purpose
DIAGNOSTIC
Masking
NONE
Enrollment
764
Participants will follow a 5-minute learning video (intervention) on CRIS and later for JNET.
Participants will follow a 5-minute learning video (intervention) on JNET and later for CRIS.
UZ Gent
Ghent, Belgium
Diagnostic Accuracy of CRIS Classification for Submucosal Invasive Carcinoma (Sensitivity, Specificity, and Overall Accuracy)
Sensitivity, specificity, and overall diagnostic accuracy of the CRIS classification for predicting submucosal invasive carcinoma compared with histopathological evaluation (reference standard). Accuracy is calculated as the proportion of correct classifications (true positives + true negatives) divided by total assessments. Results will be reported with 95% confidence intervals.
Time frame: Immediately after 5-minute training and rating of 32 images (approximately 1 hour per participant)
Diagnostic Accuracy of CRIS Versus JNET Among Expert Endoscopists (Sensitivity, Specificity, Overall Accuracy)
Comparison of sensitivity, specificity, and overall diagnostic accuracy between CRIS and JNET classifications among 12 expert endoscopists rating 32 polyp images. Accuracy calculated as proportion of correct histopathology predictions. Results reported with 95% confidence intervals.
Time frame: At completion of expert image rating (approximately 30 minutes per expert)
Diagnostic Accuracy of CRIS Versus JNET Across All Participant Categories (Sensitivity, Specificity, Overall Accuracy)
Comparison of sensitivity, specificity, and overall diagnostic accuracy between CRIS and JNET classifications across all participant categories (experts, consultants, trainees, medical students, endoscopy nurses). Accuracy calculated as proportion of correct histopathology predictions using generalized linear mixed model analysis.
Time frame: Through study completion, an average of 3 years
Change in Diagnostic Accuracy From Baseline to 3-Month Follow-up for CRIS Versus JNET (Sensitivity, Specificity, Overall Accuracy)
Change in sensitivity, specificity, and overall diagnostic accuracy from immediate post-training assessment to 3-month delayed assessment for participants using CRIS versus JNET. Reported as absolute change in accuracy with 95% confidence intervals. A smaller decrease indicates better retention of classification skills.
Time frame: 3 months after initial assessment
Inter-observer Agreement for Polyp Classification Using CRIS Versus JNET (Fleiss' Kappa Coefficient)
Inter-observer agreement among non-expert endoscopists for polyp classification using CRIS versus JNET, measured using Fleiss' kappa coefficient. Values interpreted as: \<0.20 poor, 0.21-0.40 fair, 0.41-0.60 moderate, 0.61-0.80 substantial, 0.81-1.00 almost perfect agreement.
Time frame: At completion of baseline image assessment (approximately 1 hour per participant)
Correlation Between Proposed Endoscopic Treatment and CRIS/JNET Classification (Percentage Agreement)
Percentage of cases where the proposed endoscopic treatment (piecemeal EMR, en-bloc EMR/ESD, or surgical referral) aligns with the guideline-recommended treatment based on CRIS or JNET classification. Higher agreement indicates the classification effectively guides treatment selection.
Time frame: At completion of baseline image/video assessment (approximately 1 hour per participant)
Change in CRIS/JNET Diagnostic Accuracy Following 5-Minute Structured Learning Intervention (Pre-Post Difference in Overall Accuracy)
Change in overall diagnostic accuracy from pre-intervention baseline to immediately post-intervention for CRIS and JNET classifications. Reported as absolute difference in accuracy percentage with 95% confidence intervals. The CRIS/JNET classification uses a scale where accuracy ranges from 0% (no correct classifications) to 100% (all classifications correct), with higher scores indicating better diagnostic performance.
Time frame: Immediately before and immediately after 5-minute learning video intervention (within a single 1-hour session)
This platform is for informational purposes only and does not constitute medical advice. Always consult a qualified healthcare professional.