DIFFERENTIALLY EXPRESSED GENE SIGNATURES OF ULCERATIVE COLITIS DETERMINE NOVEL REGULATORS AND PREDICT RESPONSE TO BIOLOGIC THERAPY

Abstract

Abstract Disease heterogenicity among Ulcerative colitis (UC) patients and their variable responses to therapy is not well understood. Previously, using publicly available transcriptomes from colonic tissue of UC patients, we demonstrated the immune cell landscape and pathways that are common and distinct among patients and for non-responders to biologic therapy. Here, we AIM to determine differentially expressed gene signatures specific to inflamed UC tissue and for non-responders to biologic anti-TNFa and anti-a4b7 therapy in large patient cohorts. Methods: Transcriptomes from 210 UC patient samples from inflamed and matched uninflamed tissue (GSE4183, GSE14580, GSE38713, GSE107593) and from UC patients prior to and while receiving anti-TNFa (infliximab) and anti-a4b7 integrin (vedolizumab) (GSE12251, GSE73661) treatment were analyzed. Generating and analyzing differentially expressed genes (DEGs), transcriptional signatures, and hierarchical clustering were performed using the limma Bioconductor R package and Cluster3/JavaTree software. Disease/function analysis was performed by Ingenuity Pathway Analysis (IPA) and predicting outcomes to therapy by receiver operating characteristic (ROC) curve analysis and area under the curve (AUC). Transcripts were validated in primary UC tissue samples using qPCR. Results: In three UC cohorts, we identified a specific transcriptional signature of 100 DEGs (UC100) common among patients (Figure 1A-D). The UC100 signature was validated in an independent cohort, demonstrating the ability to distinctly separate inflamed from matched uninflamed transcriptomes (p=4.65e-08). The UC100 DEGs encode protein involved in inflammation, immunity, antimicrobial response, growth, cellular movement, metabolism (lipid, amino acid), and energy production. Among them are many DEGs with unexamined roles in UC, including PCK1, HMGCS2, ACAT1, HCAR3, LPCAT1, and LIPG. Their differential expression was further validated in primary UC tissue by qPCR. Moreover, we identified twenty DEGs in UC patients resistant to anti-TNFa and anti-a4b7 therapy (UCR20) (Figure 2A-C). The UC20R DEGs encode protein involved in inflammation, immune cell trafficking, antimicrobial responses, lipid metabolism, and mitochondrial dysfunction. Four of the top DEGs predicted resistance to both therapies with a sensitivity of 73.3% and specificity 85.7%; AUC for IGFBP5: 0.86±0.008, p=0.008; STC1: 0.83±0.09, p=0.015; VNN2: 0.81±0,10, p=0.022; and SELE: 0.77±0.11, p=0.045. Conclusion: This study demonstrates differentially expressed genes in UC common across cohorts and those with predictive power to biologic therapy outcome. This provides a platform for development of more precise diagnostic tools and personally tailored therapeutic regimens for UC patients.