Key result
DKA at T1D onset linked to distinct multi-omics signatures including elevated Bacteroides clarus.
Why the study?
Environmental factors like the gut microbiota may have distinct pathogenic roles in type 1 diabetes presenting with versus without diabetic ketoacidosis at onset, but multi-omics signatures and underlying mechanisms were uncharacterized.
Observational (n=69)
Deep learning-based multi-omics analysis identified distinct gut microbiome and metabolic signatures associated with diabetic ketoacidosis at the onset of Type 1 Diabetes.
Hypothesis-generating in new-onset T1D; prospective studies needed before any clinical application.
Introduction and Objective: Environmental factors such as the gut microbiota may play distinct roles in the pathogenesis of type 1 diabetes (T1D) presenting with or without diabetic ketoacidosis (DKA) at onset. This study aimed to employ a deep learning model to identify multi-omics signatures in patients with DKA-onset T1D and to investigate the underlying mechanisms. Methods: Newly diagnosed T1D patients (n=69) were randomly enrolled and stratified into two groups: DKA and Non-DKA, based on the presence of ketoacidosis at onset. A deep learning model, Multi-Omics Variational Autoencoders (MOVE), was utilized to assess the importance of features from gut microbiome, serum metabolome, lipidome, and proteome profiles between the groups. Key microbiota and their potential associations with metabolic or protein features were screened through model perturbation analysis. Results: The MOVE model demonstrated strong capability in processing T1D multi-omics data (loss value = 0.9403). Analysis revealed distinct multi-omics signatures between the groups. The DKA group was characterized by elevated levels of s__Bacteroides clarus, 2-Arachidonyl Glycerol ether, and Phosphatidylcholine (PC) (36:2). In contrast, the Non-DKA group featured a higher abundance of f__Burkholderiaceae, along with increased levels of D-(+)-Proline, LPC (16:0), and the protein CCL28. Integrated importance evaluation of clinical, gut microbiome, metabolome, lipidome, and proteome signatures highlighted the significant role of gut microbiota in DKA onset. Perturbation-based analysis using MOVE suggested that s__Lachnospira sp000437735 might ameliorate ketoacidosis occurrence at T1D onset by facilitating the metabolism of benzyl alcohol to benzaldehyde. Conclusion: The application of a deep learning model provides novel molecular insights and potential therapeutic targets for elucidating the heterogeneity of T1D onset and for exploring individualized prevention and treatment strategies. Disclosure H. Tan: None. T. Yue: None. Y. Ding: None. Z. Gu: None. Y. Shi: None. L. Pan: None. S. Chang: None. C. Guo: None. J. Weng: None. X. Zheng: None. Funding Foundation programs supporting included Noncommunicable Chronic Diseases-National Science and Technology Major Project (2023ZD0509100; 2023ZD0509102), the National Natural Science Foundation (8247087), and Program for Innovative Research Team of The First Affiliated Hospital of University of Science and Technology of China (CXGG02)
No takes yet. Share an insight, caveat, or question.
Tan et al. (2026) conducted an observational in Type 1 Diabetes (n=69). Diabetic ketoacidosis (DKA) at onset vs. Non-DKA at onset was evaluated on Multi-omics signatures (gut microbiome, serum metabolome, lipidome, and proteome profiles). Deep learning-based multi-omics analysis revealed distinct signatures in T1D patients with DKA at onset, including elevated s__Bacteroides clarus and specific lipids, compared to those without DKA.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: