Separating prognostic association from treatment-effect modification of a baseline mucosal inflammatory-response score in ulcerative colitis biologic trials

Abstract

Background and Aims

Baseline mucosal inflammation may be associated with outcome without modifying treatment benefit. We separated these questions for a fixed inflammatory-response score in ulcerative colitis biologic studies.

Methods

A fixed 200-gene score was calculated with platform-specific probe mapping and accession-level baseline standardization. Treatment-by-score interactions were estimated in three randomized active-versus-placebo datasets. Prognostic associations were synthesized across seven treated cohorts, longitudinal change across six paired cohorts, and cellular localization using donor-aware single-cell summaries.

Results

The randomized branch included 681 participants. All three interaction odds ratios were below one (0.64, 0.46, and 0.76), but their 95% confidence intervals were wide and crossed one. Among 541 actively treated participants, higher baseline scores were associated with poorer outcome (pooled odds ratio 0.44, 0.33-0.58). Among 154 paired participants, responders had a lower baseline-adjusted follow-up score (-1.07 standardized units, -1.61 to -0.53). Analysis of 212,623 cells localized the programme across epithelial, myeloid, and fibroblast/pericyte compartments. Baseline compartment analyses included only 18-19 donors and did not establish a remission association.

Conclusions

Across the analysed public datasets, baseline mucosal inflammation showed a consistent adverse prognostic association and response-associated decline. The randomized evidence did not establish treatment-effect modification, and prospective independent validation is required before clinical use.