Only 3 of 12 gastroenterologists in a community practice were able to assess colon polyps 5 mm or less in size with 90% accuracy after completing a computer-based self-training program for "optical biopsy" using narrow band imaging, Dr. Uri Ladabaum and his colleagues reported in the January issue of Gastroenterology.

This performance was not as good as that reported for experts evaluated in clinical studies, but it was similar to that reported for students at academic medical centers.

Video Source: American Gastroenterological Association YouTube channel

Although it is "promising" that three gastroenterologists learned how to perform optical biopsy using this method, "better results in community practice must be achieved before NBI [narrow band imaging]-based optical biopsy methods can be used routinely to evaluate polyps," wrote Dr. Ladabaum of Stanford (Calif.) University and his associates (Gastroenterology 2013;144:81-91).

The American Society of Gastrointestinal Endoscopy recommends a negative predictive value threshold of 90% or greater for adenomatous histology of diminutive colorectal polyps using optical biopsy.

The investigators evaluated a self-training program that was designed to be practical to implement in a busy community gastroenterology practice. None of the study subjects had significant experience or formal training with NBI, which uses narrow band light filters to highlight colonic mucosal architecture and vasculature.

Participants first completed three self-administered units at their own pace during the first week of the program: a pretest, a learning module instructing them on using narrow band imaging to help distinguish adenomas from hyperplastic polyps, and a posttest. The two tests featured different sets of 25 endoscopic images of polyps and asked the study subjects to judge whether they were adenomas or hyperplastic, or to indicate that they couldn’t decide.

Next, the participants put their learning into practice, using optical biopsy with NBI on all colonoscopies they performed in which at least one polyp was removed. They noted the location, size, and morphology of each lesion; took photographs under white light and under NBI; classified the features they observed; predicted whether the lesions would prove to be adenomas or hyperplastic on pathologic examination; and recorded their level of confidence (high or low) in their predictions.

After the specimens were examined and classified by fellowship-trained GI pathologists, the study gastroenterologists were required to record the actual diagnosis for each lesion and encouraged to compare that with their own optical diagnoses. They received ongoing confidential feedback as to the accuracy of their work every 1-2 weeks.

Polyps were classified by size as diminutive (5 mm or smaller), small (6-9 mm), or large (10 mm or larger). The primary analysis focused on the study subjects’ success in optical diagnosis of diminutive polyps only, and each subject’s final evaluation was performed after he or she assessed at least 90 such lesions.

The 12 gastroenterologists who participated in both the ex vivo and in vivo study phases of the training program performed a total of 1,673 colonoscopies and removed 2,596 polyps. This included 1,858 diminutive, 547 small, and 177 large polyps; the size was not recorded for the remaining 14 polyps.

All 12 study subjects scored 90% or greater accuracy on the posttests they took after training.

However, only three of them performed that well in their real world practice. The other nine gastroenterologists did not achieve 90% or greater accuracy in assessing the histology of diminutive polyps using NBI.

The participants did not show typical learning curves with the technique over time. "There was no clear pattern of early learning with later stabilization of performance at a higher level," Dr. Ladabaum and his colleagues wrote.

Many factors that might have influenced subjects’ accuracy in distinguishing adenomas from hyperplasia actually proved to have no such effect. The location of the polyp within the colon, the endoscopists’ years in practice and colonoscopy volume, and the changes in subjects’ scores from pretest to posttest all showed no association with their final accuracy in optical biopsy.

One factor that predicted such accuracy was the participant’s confidence in each of his or her predictions.

Finally, there was some attrition over the course of the study, as some subjects who took part in the initial computer-based training did not proceed to the in vivo phase, "and not all participants remained engaged in the in vivo phase. Incentives may need to be developed for busy clinicians to learn and use optical biopsy techniques," the investigators wrote.

Overall, the study findings show that "high accuracy can be achieved ex vivo by community gastroenterologists after self-paced training with a computerized module." However, it remains to be seen whether a gastroenterologist’s ultimate proficiency in optical diagnosis "is determined by technical issues, dedication, motivation, or innate skill," they added.

This study was supported by Stanford University. One author reported receiving research support from and serving on the speakers bureau for Olympus Corp.

Poor performance seen after optical biopsy training