A multi-omics integration approach relying on circulating factors does not discern subtypes of childhood type 1 diabetes

Background: Type 1 Diabetes (T1D) exhibits considerable heterogeneity, impacting prediction, prevention, diagnosis, and treatment. Precision medicine aims to tailor treatments using ‘endotypes’—subtypes of disease with distinct pathophysiological mechanisms. However, proposed endotypes often lack mechanistic associations with clinical outcomes for accurately identifying T1D cases. Methods: This study introduces an approach leveraging the multi-omics factor analysis (MOFA) strategy, a computational method for unsupervised integration analysis, to explore endotypes. Analyzing data from 146 new-onset children with T1D (54 females, 92 males; age range 5–18 years), including circulating immunome, transcriptome, and serum metabolic hormones, we identify 12 factors explaining variability across the three data sets. Results: Here we find no associations, either direct or through clustering, between these 12 factors and clinical parameters, genetic predisposition, or disease outcome. These results suggest that a combination of clinical phenotypes might be responsible for the differences across T1D cases. Conclusions: These findings challenge the assumption that T1D heterogeneity reflects diverse developmental mechanisms. These results add to the ongoing debate on endotypes and carry important implications for clinical trial design—particularly in how treatments are evaluated for their effectiveness across broad and diverse patient populations.