Multi-omics reveals microbiota, metabolite, and immunological heterogeneity of age-related endotypes in type 1 diabetes.

Type 1 diabetes (T1D) exhibits age-related heterogeneity in clinical progression and immune pathology, yet the underlying molecular mechanisms remain poorly understood. Here, we integrate microbiome, metabolome, lipidome, and transcriptome profiling from 108 newly diagnosed pediatric patients with T1D, along with 56 healthy controls, to investigate age-related endotypes. Patients were stratified into early-onset (E-T1D, <7 years), intermediate-onset (I-T1D, 7-12 years), and late-onset (L-T1D, ≥13 years) groups. Multi-omics analyses revealed distinct molecular signatures among T1D subgroups. The most enriched microbial signatures were the genus Acetatifactor in E-T1D, the phylum Firmicutes A in I-T1D, and the family Bacteroidaceae in L-T1D (Linear Discriminant Analysis scores = 3.49, 5.56, and 5.78, respectively). For metabolites, pipecolic acid increased most in E-T1D, testosterone in I-T1D, while N-acetylhomocitrulline was most enriched in L-T1D. Lipidomic profiling revealed subgroup-specific alterations, with increased levels of LPA(16:1) in E-T1D, TG(16:0/18:2/18:3) in I-T1D, and TG(18:0/18:1/18:1) in L-T1D. The proportion of peripheral B cells to total lymphocytes was the highest in E-T1D (median = 11.64%) and associated with upregulated immune-related pathways, lowest in L-T1D (median = 5.99%) and linked to metabolic processes, while I-T1D (median = 8.47%) exhibited intermediate features of both groups. Integration of multi-omics interaction networks and experimental validation revealed that the microbial species Dialister invisus may promote peripheral B cell proliferation via docosapentaenoic acid, potentially contributing to early-onset T1D. Together, these findings provide a molecular framework for understanding age-related T1D endotypes and suggest potential targets for precision intervention. Workflow and key findings of the study.A multi-omics integration strategy was applied to newly diagnosed pediatric type 1 diabetes (T1D) patients stratified by age at diagnosis: early-onset (E-T1D), intermediate-onset (I-T1D), and late-onset (L-T1D), to delineate age-related T1D endotypes. Comprehensive profiling included gut microbiome, serum metabolome, lipidome, and peripheral immune transcriptome analyses. An integrated multi-omics interaction network revealed 665 direct microbiota-gene connections and 2,608 microbiota-metabolite/lipid-gene triadic interactions, highlighting a D. invisus-docosapentaenoic acid (DPA)-STMN1 axis mediating B-cell activation in early-onset T1D.