PTH-011 Defining the optimal i-scan settings for small colonic polyp characterisation - results from a large prospective series

Introduction Characterisation of small colonic polyps has been identified as a key goal for advanced imaging modalities by the ASGE and ESGE. Pentax i-Scan is an endoscopic digital contrast technology shown to be an accurate tool for characterising small colonic polyps in-vivo. I-Scan combines features which accentuate dark-light borders (surface and contrast enhancement, SE/CE) and suppression of red light whist enhancing blue/green light (tone enhancement, TE). The optimal settings for use of i-Scan in the colon have yet to be determined. Method High quality digital images of 100 small colonic polyps were recorded as part of a prospective study (NCT 01761279). Images of each polyp were recorded in 3 i-Scan modes: i-Scan 1 (SE/CE), i-Scan 2 (TE) and i-Scan 3 (SE/CE/TE). Randomised images were viewed by 2 blinded endoscopists who rated visibility of diagnostic features and predicted polyp histology (neoplastic vs non-neoplastic). Results The proportion of adenomas rated as having visible pericryptal vessels were 59%, 81% and 82% for i-Scan 1, 2 and 3 respectively. The differences between i-Scan 1 and 2 (P = 0.001) and i-Scan 1 and 3 (P < 0.001) were statistically significant. The difference in proportion of adenomas with no visible surface pattern between i-Scan 1 and 2 was statistically significant (26% vs 8%, P = 0.001), as was the difference between i-Scan 1 and 3 (26% vs 9%, P = 0.003). There was no significant difference between i-Scan 2 and 3 (8% vs 9%, P = 1.00). Mean sensitivity for adenomatous histology between the two endoscopists was 69% for polyps viewed with i-Scan 1. Compared to i-Scan 1 assessments sensitivity was significantly higher with i-Scan 2 (86%, P = 0.006 for comparison), and with i-Scan 3 (87%, P = 0.003 for comparison). There was no significant difference in sensitivity between i-Scan 2 and i-Scan 3 assessments (86% vs 87%, P = 1.000). No significant differences in specificity were found between the 3 i-Scan modes (i-Scan 1 vs i-Scan 2 P = 0.828, i-Scan 1 vs i-Scan 3 P = 1.000, i-Scan 1 vs i-Scan 3 P = 1.000). Overall accuracy increased from 79% with i-Scan 1 to 86.5% with i-Scan 2 and 87.5% with i-Scan 3. There was no significant difference in overall accuracy between i-Scan 1 and i-Scan 2 (P = 0.063) or i-Scan 2 and i-Scan 3 (P = 0.882). However there was a significant difference between i-Scan 1 and i-Scan 3 assessments (P = 0.032). Abstract PTH-011 Table 1 Diagnostic Accuracy i-Scan 1 (SE+CE) i-Scan 2 (TE) i-Scan 3 (SE+CE+TE)