Characteristics of U-BIOPRED endotypes from multi-omics integration of 24 high-dimensional/throughput data

<bold>Introduction/aim:</bold> Asthma is a complex heterogeneous disease. We propose that the integration of multi-omics and clinical data will provide a more granular definition of endotypes. <bold>Methods:</bold> Multiple omics, biomarkers, and deep clinical data were measured on 620 healthy, mild-moderate (MMA), and severe (SA) asthmatics. Healthy controls were excluded from the clustering; multi-omics associated clusters (MACs; MAC1 n=159, MAC2 n=210, MAC3 n=129) generated by integrating 24 high-dimensional data platforms. Datasets were aligned by subjects, and consensus clustering with missing labels was applied¹. MACs were characterised by atopy, biomarkers, clinical data, exacerbations, medication, questionnaires, asthma severity, and medical history. An app was developed to show group differences: https://ubiopred-handprint.dsi.ic.ac.uk/. <bold>Results:</bold> MAC1 had the most severe asthmatics (91%), followed by MAC2 (87%) and MAC3 (64%). MAC1 also had the highest sputum eosinophils (16%) and neutrophils (67%), and the highest airflow obstruction (FEV1:67% predicted). MAC1 and MAC2 shared similar clinical severity and sputum eosinophilia, but MAC2 was mostly female with lower neutrophilia. MAC3 had least sputum eosinophils (3.6%), neutrophils (45%), and airflow obstruction (FEV1:75% predicted) with lowest exacerbations. MAC1 showed the highest serum levels of IL18, IL1a. MMP3, CCL22, eotaxin, IP10, CCL18 and h-CRP, illustrating both T2 and non-T2 pathways. <bold>Discussion:</bold> Multi-omics integration of molecular and clinical data may indicate pathways to data-driven personalised medicine. The number and type of omics may, however, need careful assessment. MMA in MAC1 and MAC2 have similar molecular phenotypic profile as SA.