TOPLINE:

In patients with moderate-to-severe ulcerative colitis (UC), an artificial intelligence (AI)-based endoscopic scoring system aligned more closely with symptoms and health-related quality of life (HR-QOL) than conventional endoscopic measures and revealed clearer therapeutic benefits of ozanimod over placebo.

METHODOLOGY:

• In UC trials, endoscopic remission status, as measured by the Mayo endoscopic score (MES), and symptomatic remission status, as measured by the partial Mayo score (PMS), commonly disagree, and neither measure patients’ HR-QOL.
• Researchers sought to determine whether an AI-based endoscopic score would align more closely with UC symptomatic measures and HR-QOL than the MES. The AI model analyzes endoscopic video for spatial quantification of mucosal injury to create an overall cumulative disease score (CDS) that ranges from 0-450.
• Both the MES and CDS were derived from end-of-maintenance colonoscopy videos from 387 patients (mean age, 42.3 years) with moderate-to-severe UC who participated in the 52-week phase 3 True North trial comparing the immunomodulatory drug ozanimod with placebo.
• The AI-generated CDS and the standard MES categories (0-1 for inactive UC and 2-3 for active UC) were compared for how well they aligned with symptomatic remission per PMS and HR-QOL, measured using EuroQol Five-Dimension domains. Agreement was assessed using Cohen kappa, with higher values reflecting stronger alignment.
• Results were also examined by treatment arm (ozanimod vs placebo).

TAKEAWAY:

• Among patients labeled active by the MES (score, 2-3), those in PMS symptomatic remission had a much lower CDS than those not in remission (83.8 vs 186.9; P < .0001).
• A CDS threshold of ≤ 40 aligned better with PMS symptomatic remission than MES 0-1 (Cohen kappa, 0.72 vs 0.57) and yielded fewer cases of disagreement (14.2% vs 20.2%).
• The CDS aligned more closely with QOL than the MES, as shown by a higher agreement across quartiles of the EuroQol Five-Dimension visual analogue scale (Cohen kappa, 0.53 vs 0.44).
• Defining remission as PMS symptomatic remission plus either MES 0-1 or CDS ≤ 40 widened the difference in the percentage of patients achieving complete remission between the ozanimod and placebo groups compared with the conventional definition combining MES- and PMS-based endpoints.

IN PRACTICE:

“As AI-assisted methods allow for more granular assessment of disease activity, these methods can further hone our concepts of therapeutic efficacy in IBD [inflammatory bowel disease] clinical trials. Apart from complete disease resolution for all patients, the CDS may also help define adequate response in UC that accounts for symptoms, quality of life, and functionality,” the authors wrote.

SOURCE:

This study, led by Ryan W. Stidham, MD, MS, University of Michigan in Ann Arbor, Michigan, was published online on November 17 in the American Journal of Gastroenterology.

LIMITATIONS:

The CDS did not capture additional IBD symptoms such as abdominal pain, bowel movement at night, or fecal urgency. The durability of remission defined by low CDS was uncertain. The study did not compare histologic endpoints with CDS.

DISCLOSURES:

This research received funding from Bristol Myers Squibb, which manufactures ozanimod. One author reported serving as a consultant or on advisory boards of several pharmaceutical and healthcare companies including AbbVie, Bristol Myers Squibb, and Pfizer. Two authors reported holding intellectual property and equity on medical imaging and endoscopic analysis technologies licensed to companies by the University of Michigan. Two authors reported being the employees of the funding agency.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.