The International Blue-Light Imaging for Barrett’s Neoplasia Classification

Illustration of Barrettís esophagus.
BLI is a promising tool for optical diagnosis in Barrett's, although further studies need to be undertaken real-time in a general Barrett's surveillance population.

Using blue-light imaging (BLI), researchers developed and validated the Blue Light Imaging for Barrett’s Neoplasia Classification (BLINC) which has a 95.7% degree of sensitivity for the optical diagnosis of Barrett’s neoplasia, according to a study published in Gastrointestinal Endoscopy.

The detection of subtle Barrett’s neoplasia via surveillance endoscopy can be challenging, and the Seattle biopsy protocol can often miss focal neoplasia. BLI, an advanced form of endoscopy technology with high-intensity contrast imaging, may improve identification efforts. In this prospective noninterventional image-based study, investigators sought to develop and validate the first classification to enable characterization of neoplastic and non-neoplastic Barrett’s esophagus using BLI.

Related Articles

In phase 1 of the study, investigators identified descriptors to form the classification BLINC. Phase 2 involved 10 expert endoscopists validating the component criteria by assessing 50 BLI images. Phase 3 involved the development of a web-based training module to allow 15 general endoscopists to use BLINC. In phase 4, these endoscopists validated the classification system via an image assessment exercise, after which their pre- and post-training module results were compared.

In phase 1, the descriptors identified concerned 3 main domains with 2 to 3 descriptors each. The domains were color, pits, and vessels, and neoplasia was characterized by images containing an abnormality in at least one of these domains (distortion in mucosal pits or vessels, or color change). In phase 2, BLINC showed an overall sensitivity of 96%, a specificity of 94.4%, and an accuracy of 95.2%. All subcategories were useful for neoplasia identification, but irregular pit distribution demonstrated the highest sensitivity (97.20; 95% CI, 94.32-98.87) and accuracy (95.40; 95% CI, 93.18-97.06).

In phase 3, 15 general endoscopists completed the training module. In phase 4, 750 observations were analyzed (15 practitioners assessing 50 images), showing a significantly higher proportion of high confidence predictions after BLINC training (81.5% or 611/750) compared with before training (58.8% or 441/750; P <.0001). Although the neoplasia diagnosis sensitivity improved significantly after BLINC training (95.7%; P <.001), specificity decreased from 88.3% to 80.8% (P =.006). Before training, the neoplasia diagnostic agreement among all endoscopists was moderate (0.60; 95% CI, 0.573-0.627), and this improved to good agreement after training (K=0.67; 95% CI, 0.646-0.700), although this change was not statistically significant (P =.20).

Study limitations include a lack of a real-time in-vivo validation phase, an unrealistic concentration of neoplasia which not reflect the reality of surveillance in a community-based setting, a lack of images of low-grade dysplasia, and potential availability issues for BLI for endoscopists.

Study investigators conclude, “[W]e have developed a new classification system (BLINC) for the optical diagnosis of Barrett’s neoplasia using BLI . We also designed and implemented an online training tool on BLINC and showed that this new classification can be used effectively by both experts and nonexperts. BLI is a promising tool for optical diagnosis in Barrett’s, although further studies need to be undertaken real-time in a general Barrett’s surveillance population.”

Reference

Subramaniam S, Kandiah K, Schoon E, et al. Development and validation of the international Blue Light Imaging for Barrett’s Neoplasia Classification [published online October 3, 2019]. Gastrointest Endosc. doi: 10.1016/j.gie.2019.09.035