Endoscopy 2015; 47(03): 188-189
DOI: 10.1055/s-0034-1391374
© Georg Thieme Verlag KG Stuttgart · New York

Optical diagnosis of colorectal polyps: can this be learned by everyone?

Ralf Kiesslich
Internal Medicine Clinic II, HSK, Dr. Horst Schmidt Kliniken GmbH, Wiesbaden, Germany
› Institutsangaben
Weitere Informationen


25. Februar 2015 (online)

Screening and surveillance colonoscopy has proven to be effective in reducing the number of deaths from colorectal cancer. However, it can be clearly shown that colonoscopy is not perfect, and in fact, right-sided colonic cancers cannot be significantly reduced by screening colonoscopy [1]. Precancerous conditions such as adenomas or serrated lesions can also be overlooked and can lead to interval cancers [2]. There is a strong correlation between the adenoma detection rate and the occurrence of interval cancers. This has led to the recommendation that the adenoma detection rate be used as a surrogate marker for quality in colonoscopy [3].

Adenoma detection rate should exceed 20 %, or even better 30 %, per examiner [4]. Numerous attempts have been made to improve the efficiency of colonoscopy. Adenoma detection rates can be increased by, for example, better preparation of the patient, changing the position of the patient during withdrawal, using distal balloons or caps, using new optics, using high-definition endoscopes, or prolonging the withdrawal time [3] [5]. However, the most important factor is the examiner and intensive education is important for the performance of high-quality colonoscopy.

But focusing on high adenoma detection rates has created a new challenge. Multiple, often small polyps are now diagnosed during colonoscopy. Their malignant potential seems to be low and complete resection of all colonic polyps might be cumbersome, expensive, and unnecessary. Therefore, the educational goals for colonoscopists are to achieve high adenoma detection rates and to appropriately characterize colorectal lesions (non-neoplastic versus neoplastic).

In the past chromoendoscopy was often used in a targeted manner to characterize lesions based on crypt architecture. Japanese colleagues have shown impressively how accurately histology can be predicted using Kudo’s pit pattern classification [6]. Chromoendoscopy has also improved our knowledge of flat and depressed lesions, which can harbor a high malignant potential.

Virtual chromoendoscopy using narrow-band imaging (NBI) – or other techniques like i-scan or Fujinon intelligent chromoendoscopy (FICE) – has almost replaced standard chromoendoscopy. Virtual chromoendoscopy is simple, quick, can be switched on and off during the procedure, and is widely available. NBI is of no proven benefit in the detection of adenomas but it can be used to characterize lesions based on vessel architecture according to the NBI International Colorectal Endoscopic (NICE) classification [7].

In this issue of Endoscopy, McGill and colleagues [8] describe their randomized controlled trial to evaluate the efficiency of training for NBI-based characterization of diminutive polyps. The trial was randomized based on the availability of near-focus images from the colonoscope, which provide a more detailed view of the surface architecture. However, near focus did not lead to better characterization of colonic polyps. The trial covered two time periods with the five examiners receiving standardized training in optical analysis before starting the trial and also after the first time period ended.

NBI alone was able to predict histology with high negative predictive value (NPV). Furthermore, the screening interval could accurately be determined immediately after colonoscopy based on the optical diagnosis of the detected polyps. The authors stress the fact that NBI-based optical diagnosis can easily be learned and that high diagnostic accuracy could be sustained over both time periods of the randomized trial.

The question arises whether this study will be sufficient to recommend the so-called “resect and discard” strategy for diminutive polyps. The American Society of Gastrointestinal Endoscopy (ASGE) has made a clear recommendation saying that the NPV has to exceed 90 % and the prediction of the surveillance interval should be correct in more than 90 % of the cases. These data should always be achieved [9].

Therefore, we have to carefully analyze the results of the current study on a per examiner view. Two out of five endoscopists failed to achieve the NPV threshold in the close-up view arm (one in the first half, the other in the second half of the study). Three out of the five endoscopists failed in the standard arm (two in the first half of the study, one in the second half).

Whilst the overall results of this study are impressive, they use pooled data for the five endoscopists and we must consider each individual examiner. The single endoscopist analysis is also relevant for adenoma detection rates because these individual results translate into the clinical outcome for the cohort of examined patients.

I do not believe that this study will immediately lead to the wide acceptance of the “resect and discard” strategy. But I do believe this is a highly relevant study because it shows convincingly that continuous and repeated training in optical diagnosis has long-term impact on the quality of colonoscopy. The adenoma detection rate is an accepted surrogate marker for quality of colonoscopy (reduction of interval cancers). Optical diagnosis of colorectal polyps using surface characteristics and reporting NPVs and surveillance intervals (reducing the need for histological diagnosis) may also evolve as a quality marker.

Future studies involving more endoscopists and evaluating over a longer time frame are needed before resect and discard will enter routine clinical practice. However, the way has been paved.