Colorectal polyp size measurement: can we trust our own eyes?

Polyp size measurement permeates every aspect of colonoscopy. Our entire body of literature is based on the simple concept of estimating the size of these polyps. We are told that polyps ≥10 mm require close follow-up. We are told to use cold snares for polyps <10 mm and hot snares for polyps ≥10 mm. We are told that forceps should be limited to 1–3-mm polyps; that the risk of advanced pathology in 1–5-mm polyps is low, therefore we can discard these polyps without histologic assessment. But what are all these recommendations based on? We assume that these strict size cutoffs are golden rules to follow; however, for such a simple and important pillar of endoscopy, polyp size measurement has been unsystematic and only a second thought in our practice until recently. When a true 10-mm polyp is encountered, there is currently no guarantee that two endoscopists would agree on its size, and the same patient could receive vastly different management as a consequence.

In this issue of Endoscopy, van Bokhorst et al. shed more light on this phenomenon of variability in polyp size measurement [1]. The authors used an endoscope with a novel virtual scale function (SCALE EYE; Fujifilm, Tokyo, Japan) or snares to measure polyps, and video recordings were shown to 17 endoscopists for assisted and unassisted measurement. The authors found that 40%–50% of polyps were assigned to different size categories when using visual or snare-based measurement. Additionally, variance in size measurement was significantly lower when a virtual scale was used compared with snares or visual assessment (0.58 vs. 2.02 vs. 2.06) and this remained true for both experts (P < 0.001) and trainees (P < 0.001). What is interesting is the extent of variability between experts when not using calibrated tools for measurements, with one expert classifying 3 polyps as ≥10 mm at one extreme, and another classifying 15 polyps as ≥10 mm at the other extreme.

„Should we repeat landmark studies for more accurate assessment of outcomes within the prespecified size categories or is the current approximate and variable measurement paradigm “good enough” to guide our practice?“

These findings underscore a worrisome realization for the current body of literature evaluating all aspects of endoscopic practice: that the polyp size included in every study is not to be trusted. Indeed, 6–7-mm even 8-mm polyps have likely been included in both resect-and-discard studies for 1–5-mm polyps and studies evaluating resection techniques for 10–19-mm polyps. Endoscopic mucosal resection trials have likely included 17–19-mm polyps. Studies assessing the risk of ≥10-mm polyps at follow-up have likely included a significant number of polyps <10 mm. Tools to reduce variability in polyp size measurement are therefore of paramount importance if any recommendations are to be made in the future based on polyp size.

Recent advancements in endoscopic size measurement have shown how truly variable this measurement can be, with studies showing very low accuracy when performing visual or snare-based size measurement, which persists among experts, trainees, gastroenterologists, and surgeons [2]. Advancements such as virtual scales have been proposed to standardize polyp management by ensuring better agreement for size measurement [3]. We found in our own randomized controlled trial that such a tool could improve size categorization when using caliper measurements in polyps immediately after resection and before formalin fixation as a gold standard [4]. Measurements in pathology were also highly inaccurate due to formalin shrinkage [5].

Artificial intelligence (AI) could provide a definite solution to the problem of polyp sizing, in contrast to polyp detection and diagnosis where we have seen challenges in the implementation of AI technology, as well as unclear efficacy. However, although successful in reducing variability, virtual scales are operator dependent. Additionally, the laser needs to be placed accurately next to the polyp at the right distance to ensure accurate scale projection on the screen. These scales also measure polyps on the right–left axis but are not useful for measurement along the oral–anal axis, as the depth on which the scale is projected is fixed on the point of laser contact with the colonic mucosa. An operator-independent AI solution would be very beneficial in removing the variables involved in accurate measurement [6] [7]. Such tools are currently in development and larger trials evaluating their efficacy will be available in the not-too-distant future.

Once accurate tools for polyp measurement become routinely available with standardized and granular polyp sizing information, will we find out that 8–9-mm polyps are just as high risk as ≥10-mm polyps? Or that the cutoff should be increased to 12 mm instead? Could we consider 6–7-mm polyps in the diminutive size category, or should we reduce the cutoff to 1–3 mm? Should we repeat landmark studies for more accurate assessment of outcomes within the prespecified size categories or is the current approximate and variable measurement paradigm “good enough” to guide our practice? With their study, van Bokhorst et al. point to a re-evaluation of our combined published literature as the avenue going forward.

Publication History

Article published online:
16 September 2024

• References

• 1 van Bokhorst QNE, Houwen BBSL, Hazewinkel Y. et al. Polyp size measurement during colonoscopy using a virtual scale: variability and systematic differences. Endoscopy 2024; 56
• 2 Djinbachian R, Popescu I Crainic, Pioche M. et al. Accuracy in polyp size measurement among surgeons, gastroenterologists, trainees, and experts: a prospective video-based study. Am J Gastroenterol 2024; 119: 532-538
• 3 von Renteln D, Djinbachian R, Zarandi-Nowroozi M. et al. Measuring size of smaller colorectal polyps using a virtual scale function during endoscopies. Gut 2023; 72: 417
• 4 Taghiakbari M, Djinbachian R, Haumesser C. et al. Measuring size of colorectal polyps using a virtual scale endoscope or visual assessment: a randomized controlled trial. Am J Gastroenterol 2024; 119: 1309-1317
• 5 Djinbachian R, Khellaf A, Noyon B. et al. Accuracy of measuring colorectal polyp size in pathology: a prospective study. Gut 2023; 72: 2015-2018
• 6 Abdelrahim M, Saiga H, Maeda N. et al. Automated sizing of colorectal polyps using computer vision. Gut 2022; 71: 7
• 7 Sudarevic B, Sodmann P, Kafetzis I. et al. Artificial intelligence-based polyp size measurement in gastrointestinal endoscopy using the auxiliary waterjet as a reference. Endoscopy 2023; 55: 871-876