Introduction The course of Crohn’s disease (CD) varies substantially between affected individuals, but reliable prognostic markers are not available in clinical practice. This hinders disease management because patients with aggressive disease will be undertreated by conventional “step-up” therapy, while those with indolent disease would be exposed to the risks of unnecessary immunosuppression of a “top-down” approach. Previously, we have described a transcriptional signature that is detectable within peripheral blood CD8 T cells at diagnosis and which correlates with subsequent disease course. To translate this work to the bedside and overcome the technical challenges of separating cell populations, we sought to develop a whole blood qPCR-based biomarker that can re-capitulate the CD8 subgroups without the need for cell separation. Here we describe the development and validation of this biomarker and the upcoming biomarker-stratified trial that will test whether it can deliver personalised medicine in CD.

Method From a training cohort of 69 newly diagnosed IBD patients, we simultaneously obtained a whole blood PAXgene RNA tube and peripheral blood CD8 T cell sample. Gene expression in both samples was measured by microarray. After confirming that the CD8 transcriptional signature was detectable and correlated with prognosis, we used machine learning to identify a transcriptional classifier in whole blood gene expression data that would re-capitulate the CD8 transcriptional subgroups. Model selection was performed using Bayesian Information Criterion and the genes identified were subsequently tested by qPCR and optimised to produce an 18 gene qPCR assay.

Results Independent validation of this biomarker was established using a second, independent cohort of 85 newly diagnosed patients with CD from 4 sites around the United Kingdom. This validated the biomarker and confirmed that the subgroups it identified had significantly different disease courses (analogous to those observed with the CD8 T cell subgroups). The hazard ratio for time to treatment escalation in this validation cohort was 3.52 (1.84–6.76, 95% confidence intervals, p=0.0002). We now propose to conduct the first ever biomarker-stratified trial in any inflammatory disease to determine whether this biomarker can deliver personalised medicine in CD.

Conclusion We have developed, optimised and validated a whole blood qPCR classifier that is able to predict disease course from diagnosis in IBD patients. This represents a major step towards personalised therapy in IBD, and we will soon investigate whether this could make personalised medicine a reality in CD.

Disclosure of Interest None Declared