Subscribe to RSS
DOI: 10.1055/s-0039-1681615
RELIABILITY AND ACCURACY OF BLUE LIGHT IMAGING FOR STAGING OF INTESTINAL METAPLASIA IN STOMACH (BLIMPS)
Publication History
Publication Date:
18 March 2019 (online)
Aims:
A grading endoscopic system using high-resolution scopes with NBI has been shown to accurately identify patients with extensive GIM (EGGIM) that need surveillance. However, no description is available with alternative systems such as the new system of Fujifilm, BLI. We aim to determine the reliability and accuracy of BLI regarding the diagnosis and staging of GIM.
Methods:
A consecutive series of patients (n = 29), previously assessed by NBI, with a full spectrum of gastric changes (OLGIM 0-IV) were submitted to gastroscopies using FujifilmEG-760ZHD/ELUXEO-VP7000 and endoscopists (blinded to the previous histologic status) were asked to determine EGGIM score on real-time using BLI-bright mode (eg, 0 – 2 for the lesser curvature of antrum and corpus, greater curvature of antrum and corpus and incisura, total 0 – 10). Reliability with BLI using the previous classification by Pimentel-Nunes P et al. among 3 observers was determined with WLE, LCI and BLI (n = 32 per site images). Secondly, accuracy was determined by comparing with previously EGGIM (cutoff of 4) with NBI and current OLGIM status.
Results:
The overall interobserver reliability for histologic presumption based on endoscopic images with BLI (wK 0.80[95% CI:0.64 – 0.93]) and LCI (wK 0.76[95% CI:0.53 – 0.90]) was substantially better than WLE (wK 0.43[95% CI:0.20 – 0.66]. The proportion of certainty varied between 50 to 62% for WLE, 59 to 81% for LCI, 78 to 91% for BLI. Accuracy is reported in table 1.
EGGIM using BLI-Bright |
EGGIM using NBI |
OLGIM (n = 27) |
||
0 – 4 |
5 – 10 |
0-II |
III-IV |
|
0 – 4 |
12 (92%) |
4 |
16 (84%) |
0 |
5 – 10 |
1 |
12 (75%) |
3 (16%) |
8 (100%) |
> 90% agreement |
AUC 0.92 (CI 95%: 0.82 – 1.00) |
Conclusions:
BLI is reliable in determining the presence of GIM. BLI-bright seems to agree significantly with NBI evaluation (90% agreement) and preliminary data suggests very high sensitivity for identifying those at risk (OLGIM III/IV). External multicentre assessment is required for further validation.