AI endoscopic score shows better agreement with ulcerative colitis remission than Mayo score in RCT.
This was a phase 3 randomized controlled trial in 387 subjects with ulcerative colitis from the TrueNorth trial. The study compared an artificial intelligence endoscopic cumulative disease score (CDS) to the standard Mayo endoscopic score (MES) over a 12.0-month follow-up. The primary outcome was agreement with symptoms and health-related quality of life.
In endoscopically active subjects achieving partial Mayo symptomatic remission, the CDS was lower (83.8 vs 186.9; P < 0.0001). The agreement of EuroQol 5 Dimension with CDS was better than with MES (κ = 0.53 vs 0.44). A low CDS cutoff (<40) had better agreement with partial Mayo remission (κ = 0.57 vs 0.72; area under the curve 0.85).
The difference between ozanimod and placebo remission with adequate endoscopic criteria increased from 22.6% to 28.6%. Safety data, including adverse events, serious adverse events, discontinuations, and tolerability, were not reported.
Key limitations include the observational analysis within an RCT, where association is reported, not causation. The practice relevance is that AI-enabled endoscopic scoring has better agreement with symptomatic remission and HR-QoLs compared with MES and could quantitatively redefine adequate mucosal healing.