P0466Central reading of routine ulcerative colitis endoscopy at a tertiary IBD centre: A prospective pilot study

Abstract Background Central reading is the regulatory standard for endoscopic endpoints in ulcerative colitis (UC) trials, yet real-world evidence comparing local versus central readings is limited. Real-world populations also differ significantly from clinical trial participants. Thus, it is unclear whether trial-grade central reading would meaningfully impact daily care. Methods A prospective pilot study was conducted to evaluate the accuracy, reliability, and error patterns of routine local readings using central reading as the reference standard. We prospectively enrolled 59 consecutive UC patients (mean age 50.8 years (range 20-86), 49.3% female) undergoing colonoscopy at the Montreal University Hospital Center. Full-length colonoscopy recordings were captured with standardized video protocols. Local endoscopists prospectively assigned Mayo Endoscopic Subscores (MES) in real-time. One expert central reader then independently reviewed and scored all videos. Primary outcome was the accuracy of local MES. Secondary outcomes included the linearly weighted κ agreement measure between local and central MES, misclassification direction and amplitude, and qualitative error taxonomy. We also assessed the proportion of cases with clinically relevant read differences. Results Accuracy of local MES was 76.3% (45/59; 95% confidence interval, CI: 63.1–86.0). Agreement between local and central MES was substantial, with weighted κ = 0.636 (0.479–0.794). Among the 14 cases with local read misclassification (Table 1), 12 MES scores had a ± 1 point difference, while 2 scores had a 2-point difference. Local reader scores were less severe than central reader scores in 13 cases, most commonly downgrading MES 1 to MES 0 (6/13). Central readers identified additional ulcerations in 28.6% (4/14) of cases, significantly upgrading MES scores from 1 to 3 (in 2/4 of cases) or from 2 to 3. Correctable technical issues were identified in most discrepant cases, including inadequate mucosal washing limiting lesion visualisation and suboptimal luminal distension obscuring vascular pattern. MES discrepancies would be likely to change therapeutic interpretation in up to 50% (7/14) of cases, when making the distinction between MES 0-1 or MES 2-3 activity. Conclusion This study demonstrates the feasibility and value of central review in routine practice. Local routine UC readings demonstrated accuracy and substantial agreement with central reading, yet discrepancies were often clinically meaningful and related to correctable technical errors, primarily inadequate washing leading to missed ulcerations. Future solutions could include standardized video acquisition with AI-based UC scoring with local training and improved mucosal preparation to enhance diagnostic accuracy. Conflict of interest: Ms. Oleksiw, Megan: No conflict of interest von Renteln, Daniel: Daniel von Renteln is supported by a “Fonds de Recherche du Québec Santé” career development award. He has also received research funding from ERBE Elektromedizin GmbH, Ventage, Pendopharm, Fujifilm and Pentax, and has received consultant or speaker fees from Boston Scientific Inc., ERBE Elektromedizin GmbH, and Pendopharm. Michal, Victoire: No conflict of interest Daoud, Dane Christina: No conflict of interest Bernard, Edmond-Jean: No conflict of interest Orlicka, Katarzyna: No conflict of interest Leduc, Raymond: No conflict of interest Daoust, Louise: No conflict of interest Liu Chen Kiow, Jeremy: No conflict of interest Djinbachian, Roupen: Roupen Djinbachian has received speaker fees from Fujifilm. Deslandres, Erik: No conflict of interest Bouin, Mickael: No conflict of interest Panzini, Benoit: No conflict of interest Bouchard, Simon: No conflict of interest Battat, Robert: Speaker/consulting/moderator: Bristol Myers Squibb, Johnson and Johnson Innovative Medicine, AbbVie, Takeda, Ferring, Celltrion, Pfizer, Merck, Eli Lilly Advisory boards: Celltrion, AbbVie, Pfizer, Janssen, Bristol Myers Squibb, Takeda, Merck, Eli Lilly Educational grants and educational sponsorships: Abbvie, Johnson and Johnson Innovative Medicine, Pfizer, Eli Lilly, Amgen, Pendopharm, Kye Pharma, Merck, Celltrion, Organon, Takeda Travel support: Pfizer, Celltrion, Johnson and Johnson Innovative Medicine, Abbvie.