Data-driven subphenotyping uncovers ulcerative colitis subtype with high risk for relapse in Japan: a prospective multicenter cohort study

Background Clinical symptoms do not necessarily align with disease activity in ulcerative colitis (UC). It remains unclear whether patient-reported outcomes (PROs), particularly quality of life (QOL) measures, can help characterize clinically meaningful subphenotypes among Japanese patients with UC in remission. This study aimed to identify and characterize UC subphenotypes using clustering analysis. Methods We used data from a multicenter, prospective UC patient registry in Japan, called "YOu and Ulcerative colitis: Registry and Social network (YOURS)" (December 2018–June 2022). We assessed laboratory values and PRO data, including clinical symptoms and QOL data (e.g., fatigue, anxiety, disease-specific QOL), of patients with UC in remission. Discovery (N = 365) and replication (N = 982) datasets were independently created and subjected to clustering analysis. Data were standardized for sex (a potential confounder for various measures). Findings PRO data correlated poorly with traditional clinical laboratory parameters. Three reproducible clusters were detected in both datasets: one cluster was characterized by lower QOL and average laboratory profiles, while the other two clusters showed preserved QOL scores either with or without distinct laboratory profiles. The cluster with lower QOL and average laboratory profiles demonstrated increased relapse rates during a 3-year follow-up in each dataset compared with the respective other two clusters. Interpretation PRO assessments offer unique, clinically relevant data, distinct from routine biomarkers and clinical scores. Stratifying patients with UC using combined QOL and clinical laboratory parameters may help identify patients at higher risk of relapse and warrants further evaluation for its potential role in clinical decision-making. Funding This study was funded by Takeda Pharmaceutical Company Limited.