Document resource
Background Mucosal fibrosis and epithelial-mesenchymal transition (EMT) are increasingly recognised as drivers of treatment-refractory IBD, yet no transcriptomic tool exists to stratify patients by fibrotic risk prior to treatment. We aimed to develop and validate a Fibrosis Risk Score (FRS) from colonic biopsy transcriptomes and determine its utility in predicting anti-TNF response.Methods A 47-gene FRS was derived from established fibrosis and EMT signatures encompassing matrix metalloproteinases, collagens, TGF- β pathway components, EMT transcription factors, and fibroblast activation markers (IDDF2026-ABS-0225 Figure 6. Fibrosis risk score projected onto UMAP). Gene expression was z-scored and averaged to generate a continuous per-sample FRS across 616 colonic biopsies from four integrated GEO cohorts (GSE75214, GSE179285, GSE16879, GSE36807) following ComBat batch correction. Unsupervised consensus clustering (k=3) identified three molecular subtypes: Metabolic-Absorptive (C1), Hypoxia-Metabolic (C2), and Inflammatory-EMT (C3). FRS was validated for treatment response prediction in pre-treatment biopsies from GSE16879 (n=61) by ROC analysis.Results FRS was strongly stratified by molecular subtype: C3 exhibited significantly higher FRS (0.729±0.531) versus C2 (-0.406±0.320) and C1 (-0.286±0.466) ( IDDF2026-ABS-0225 Figure 1. FRS distribution by molecular subtype). FRS was elevated in UC (mean=0.322) relative to CD (-0.077) and controls (-0.611), with wide intra-CD variance reflecting molecular heterogeneity (IDDF2026-ABS-0225 Figure 2. FRS by clinical diagnosis).Pre-treatment FRS significantly predicted infliximab non-response (responders: -0.078±0.394 vs non-responders: 0.599±0.584; t=-5.21, p<0.0001; AUC=0.828) (IDDF2026-ABS-0225 Figure 3. ROC FRS predicts infliximab non-response). At the optimal threshold (FRS30.365), 85% of high-FRS patients were non-responders versus 29% of low-FRS patients, a 4.7-fold difference in response rates (IDDF2026-ABS-0225 Figure 4. FRS responders vs non responders). UMAP projection confirmed a continuous fibrotic gradient across transcriptomic space, independent of clinical diagnosis (IDDF2026-ABS-0225 Figure 5. Infliximab response by fibrosis risk group).Conclusions A 47-gene Transcriptomic Fibrosis Risk Score predicts anti-TNF non-response with AUC=0.828, outperforming clinical diagnosis as a treatment stratification tool. High-FRS patients, identifiable across both UC and CD, have a 4.7-fold lower infliximab response rate and may represent candidates for alternative biological therapies targeting IL-17, JAK-STAT, or fibroblast activation pathways. Prospective validation of the FRS as a pre-treatment biomarker could transform IBD treatment selection.Abstract IDDF2026-ABS-0225 Figure 1Abstract IDDF2026-ABS-0225 Figure 2Abstract IDDF2026-ABS-0225 Figure 3Abstract IDDF2026-ABS-0225 Figure 4Abstract IDDF2026-ABS-0225 Figure 5Abstract IDDF2026-ABS-0225 Figure 6