P1308 Fibrotic and barrier-loss programs define anti-integrin response in Ulcerative Colitis

Background Ulcerative colitis (UC) is a clinically heterogeneous disease with unpredictable treatment outcomes. Anti-integrin therapies, including vedolizumab and etrolizumab, have advanced management, but reliable predictors of response are lacking. Pre-treatment transcriptomic states may capture distinct immune and stromal programs that influence therapeutic outcomes. The central hypothesis of this study is that such baseline molecular profiles can stratify UC patients into biologic-responsive versus refractory subtypes, providing a foundation for precision treatment. Methods Bulk RNA-seq (GSE72819), microarray (GSE73661), and scRNA-seq (GSE228165) datasets from anti-integrin–treated UC patients were analyzed (Table 1). After normalization and batch correction, three clustering strategies were compared: Partitioning Around Medoids, Gaussian Mixture Models, and an autoencoder. Autoencoder-derived subtypes were selected1,2,3,4. Differential expression, random forest feature selection, enrichment analysis, and integration with scRNA-seq signatures to characterize cluster biology and immune programs. Single cell integration was done by using Gene Set Variation Analysis (GSVA) and Cell-type Identification By Estimating Relative Subsets Of RNA Transcripts (CIBERSORTx). Results The autoencoder achieved the highest silhouette score (0.275), defining two robust UC subtypes. Cluster 1, enriched for non-responders, showed fibroblast activation and extracellular matrix remodeling with upregulation of IL11, TWIST1, PRRX1, THBS2, ADAM12, and TNFAIP6, consistent with a fibrotic program. Cluster 2, enriched for responders, displayed epithelial barrier loss with downregulation of transporter and differentiation genes including SLC26A3, SLC4A4, CLCA4, HNF4A, CDX2, and FCGBP (Figure 1 ). Random forest identified 40 discriminative genes reinforcing subtype separation. Integration with scRNA-seq revealed immune divergence: Cluster 1 showed CD8⁺ T cell enrichment, aligning with cytotoxic activity and stromal remodeling; Cluster 2 showed CD4⁺ T cell, B cell, and NK cell activity, consistent with an immune-responsive phenotype. GSVA confirmed program-level differences, while CIBERSORTx highlighted dendritic cell enrichment in Cluster 2 with borderline monocyte/B-cell shifts. Conclusion Transcriptomic profiling stratified UC into two biologically distinct subtypes: a fibrotic, non-responder–enriched state and an inflamed but responsive state. Findings implicate IL-11/TGF-β signaling as potential targets for refractory disease and highlight anti-fibrotic strategies for the fibrotic subtype. Barrier-restorative approaches may be most effective in the responsive subtype, providing a framework for precision therapy in UC. References: 1. Love MI, Huber W, Anders S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. 2. Irizarry RA, Hobbs B, Collin F, et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003;4(2):249-264. 3. Leek JT, Johnson WE, Parker HS, Jaffe AE, Storey JD. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. 2012;28(6):882-883. doi:10.1093/bioinformatics/bts034 4. Rudy J, Valafar F. Empirical comparison of cross-platform normalization methods for gene expression data. BMC Bioinformatics. 2011;12:467. doi:10.1186/1471-2105-12-467 Conflict of interest: Dr. Santos, Daniela: The authors declare no conflicts of interest. Acharjee, Animesh: The authors declare no conflicts of interest.