Large-scale transcriptomic meta-analysis identifies constitutive and inflammation-dependent layers in ulcerative colitis

Ulcerative colitis (UC) is a chronic inflammatory bowel disease for which the vast majority of the approved therapies target the immune system. We aimed to identify consistently dysregulated genes and pathways across independent UC transcriptomic cohorts, highlighting constitutive and inflammation-dependent changes. We performed a random-effects meta-analysis of 17 microarray datasets from the Gene Expression Omnibus (1,095 mucosal biopsies, 10 platforms), comparing inflamed UC, uninflamed UC, and inflamed Crohn’s disease (CD) to controls and to each other, and validated the results in 9 independent RNA-seq cohorts. The inflamed UC analysis revealed an upregulated inflammatory transcriptomic profile in UC, providing a rationale for the use of all approved anti-inflammatory therapies. In parallel, the predominant downregulated signal was metabolic, driven by PPARGC1A, PPARGC1B, and ESRRA, indicating a coordinated collapse of mitochondrial energy metabolism. Comparing the three UC contrasts showed this collapse to be inflammation-dependent and it replicated very well on RNA-seq. Uninflamed UC mucosa was nonetheless distinct from control mucosa, carrying a suppressed glucuronidation program and persistent low-grade interferon and complement signaling, although these constitutive signals were less stable across cohorts and replicated only in part on RNA-seq. The metabolic deficit was more severe in UC than in CD, while immune pathways were shared. These findings suggest a two-layer model of UC pathology, in which the inflammation-dependent layer acts upon a mucosa that is abnormal even without visible inflammation.