Novel tool can reliably exclude submucosal invasion in colorectal polyps

Endoscopists of varying experience were able to reliably exclude submucosal invasion (SMI) in large, non-pedunculated, colorectal polyps (LNPCPs) using a novel clinical decision support tool. The blink impression only was highly predictive for the absence of SMI. The tool is easy-to-use and just requires a short training intervention. Application of this tool in clinical practice could reduce the negative consequences of missing SMI during an endoscopic assessment.

The presence of SMI is decisive for the endoscopic technique used by the specialist to treat polyps and determines whether additional systemic therapy should be initiated. The detection of SMI in LNPCPs is currently poor [1]. Therefore, the current study aimed to develop an evidence-based, easy-to-use, clinical decision support tool for endoscopists to detect SMI in colorectal polyps [2].

Dr Lynn Debels (University Hospital Ghent, Belgium) and colleagues created an algorithm which used the blink impression of the endoscopist, scanning for an overt demarcated irregular area, and scoring several parameters (size, location, morphology, Paris classification of early cancer) if no overt demarcated area was present to estimate the risk of SMI. The authors prospectively selected 20 LNPCPs with help from experts. Endoscopy videos were standardised, and a 10-minute educational video was developed to explain the use of the clinical decision tool. In total, 37 endoscopists with a broad range of experience estimated the risk of SMI in all 20 selected LNPCPs, following the algorithm. Participant observations were compared with expert opinions and histopathology.

Overall, the participating endoscopists could estimate the presence of SMI with an accuracy of 75.0%, with a negative predictive value (NPV) of 93.2% and a positive predictive value (PPV) of 70.4%. The blink impression of the participants could predict SMI with an accuracy of 72.3%. The NPV of their blink impression was 97.5%, whereas the PPV was 41.8%. The presence of an overt demarcated area was detected with an accuracy of 78.6%. Again, the NPV was higher than the PPV (97.6% and 48.2%, respectively). Expert and participant observations of covert SMI parameters showed an overlap ranging between 61.8% (morphology) and 94.8% (location). In this study, 71.3% of the cases would receive the correct treatment, whereas 13.1% would receive undertreatment (mostly high-risk patients), and 15.6% would be subjected to overtreatment.

1. Tate DJ, et al. Endosc Int Open 2020; 8(3):E445-E455
2. Debels L, et al. The accuracy of human detection of submucosal invasive cancer in large non-pedunculated colorectal polyps: analysis of 739 individual assessments of large non-pedunculated colorectal polyps using a novel clinical decision support tool. LB01, UEG Week 2021 Virtual Congress, 03–05 October.

 

Copyright ©2021 Medicom Medical Publishers



Posted on 27 October 202127 October 2021