W183 - Multimodal Analysis of IBD Patients Coupled with Machine Learning Identifies Immunometabolism Targets for Potential Therapies

Abstract Text: Inflammatory bowel disease (IBD) is a chronic immune disease with increasing global burden. Whilst different therapeutic options are available, their efficacy is often limited to only a fraction of the patients, making the identification of novel therapeutic strategies a compelling need. The etiology of IBD entails genetic and environmental factors, involving different cellular processes and pathways. Immunometabolism and the underlying adaptation of immune and stromal cells to their microenvironment, has recently emerged as potential drivers in the onset and progression of the disease.

Here, we systematically analyzed over 3000 metabolic genes and observed broad immunometabolism perturbations across bulk and single-cell transcriptomes of IBD patients. Then, we aimed to uncover new strategies to inhibit metabolic reprogramming in IBD. On this ground, we developed MetID, a novel data and analytical framework that leverages the largest integrated multi-modal data resource available in IBD, along with AI/ML methods, to identify promising immunometabolic targets with potential causative role in the etiology of ulcerative colitis (UC) and Crohn's disease (CD). By using MetID, we could surface the pathological role of lipid metabolism in fibroblast and macrophage cells in driving the disease. From our results we identified SPHK1, a sphingosine kinase which belongs to the clinically validated sphingolipid pathway, as novel therapeutic target for UC and CD and differentiated it from S1PR agonists, in silico as well as experimentally.

This work highlights the added value of multi-modal analysis to dissect this complex disease and paves the way to the discovery of novel therapeutic opportunities in IBD.