LAUSR

Introduction Anti-tumour necrosis factor (TNF) agents are used to treat UC but response is variable. Apart from concurrent immunomodulatory therapy and there are no clear predictors of efficacy. Analysis of transcriptome from peripheral blood CD4 and CD8 T-cells has been shown to predict disease outcome in inflammatory bowel diseases (IBD) but this strategy has not been tested to predict response to biological therapy. We investigated the utility of baseline CD4 and CD8 transcriptome in predicting response to anti-TNF agents in UC. Methods Patients who were commenced on any anti-TNF therapy for ambulant UC were included in this single centre cohort study. Clinical response and remission was defined using full or partial Mayo score at week 14. RNA was extracted from peripheral blood CD-4 and CD-8 populations and subjected to transcriptome analysis using human Clariom D analysis. Transcriptome Analysis Console (TAC) 4.0 from ThermoFisher Scientific was used to analyse Expression Array feature intensity (CEL) files. The analysis was carried out with the Clariom_D_Human NetAffx Library. Statistical analysis to detect differential expressed genes was carried out with default settings of TAC, except that the use of FDR p-values was set from false to true. Results Ten patients with UC with a median age of 35 (range 19–69) and median Mayo score of 8 (range 2–12) were included. Three (30%) had pancolitis and 6 (60%) of patients were on concomitant immunomodulators. At week 14, six (60%) and 4 (40%) patients achieved clinical response and remission respectively. Of the 135,750 genes tested, differential expression was noted in over 900 genes between responders and non-responders at a P value of <0.05. However, there was only one differentially expressed gene in the CD4 cell population in patients who achieved clinical remission with an FDR P-val < 0.05. There was a 25.87 fold higher expression of the major histocompatibility complex, class I, U (pseudogene) in patients who failed to achieve remission. Similarly, HLA-U pseudogene expression at baseline was higher in patients with lack of clinical response (12 fold vs 7 fold, P<0.01). Receiver operating characterestic (ROC) analysis for HLA-U pseudogene predicted clinical remission with >95% sensitivity and specificity (P<0.0001, figure 1).Abstract OWE-37 Figure 1 ROC curve of HLA-U psudogene to predict clinical remission to anti-TNF agents at week 14 Conclusions CD4 transcriptome analysis at baseline identified differentially expressed genes in patients with lack of clinical remission. This has potential utility as a novel non-invasive biomarker of resposne to anti-TNF therapy in UC. Our findings require further validation in a larger cohort.