(VP010) AN UNSUPERVISED APPROACH TO SYSTEMATICALLY ANALYSIS THE MULTI-OMICS PROFILES ASSOCIATED WITH T2D INSULIN RESISTANCE AND BETA-CELL FUNCTION: AN EU-RHAPSODY STUDY

Background: Considerable heterogeneity exists in Type 2 diabetes disease onset and subsequent outcomes. Multi-omics analysis may provide a means to better assess inter-patient variations and design personalised approaches. Previous T2D biomarker discovery (Slieker, et al. Diabetes, 2021) mainly focused on associating biomarkers with clinically pre-assigned patients (Alhqvist et al. Lancet Diabetes Endocrinol, 2018).

METHODS AND RESULTS: Here, we attempt to group T2D patients with similar multi-omics profiles using a multi-omics integrating clustering approach, Similarity Network Fusion (SNF). Each T2D subgroup’s unique omics profiles were then associated with T2D progression. We leveraged two large EU T2D cohorts, the Hoorn Diabetes Care System (DCS) and Genetics of Diabetes Audit and Research in Tayside Scotland (GoDARTS) (Slieker et al, Nature Communications, in press, 2023) with total of 1,134 subjects. Based on 180 circulating lipids and 1195 circulating proteins, two distinct subgroups can be observed in both cohorts. Two subgroups are likely to represent the different stages of T2D insulin resistance and beta-cell function with differences in HOMA2 (p=0.0008;4.2e-11;1.1e-09, -B, -IR and -S, DCS), C-peptide (p=3.7e-11;2.5e-06) and overall disease progression (HR=0.6;0.7). Examined across cohorts, a number of discriminative omics features can be observed. Immune-related proteins, such as IL-18R, CLP, IL-1R and COD antigen showed strong but differing associations with insulin resistance, which may be related to the inflammation associated with T2D. Moreover, growth factors such as GHR and IGFs also exhibited strong associations with the present study. For lipids, patients with less severe insulin resistance exhibit elevated levels of sphingomyelins.

Conclusion: In conclusion, we systematically analysed the multi-omics profiles associated with T2D insulin resistance and beta-cell function. Our data-driven, bottom-up approach allows potential novel biomarker discovery which may be masked by traditional approaches. The molecular signatures then may be used to investigate molecular mechanisms underpinning progression and provide insights for precision medicine.