Introduction and Objective: A convergence of dysregulated metabolic, polygenic and environmental factors drives the pathophysiology of diabetes mellitus. We performed unsupervised integration of transcriptomic and metabolomic data among participants in the Rare and Atypical Diabetes Network (RADIANT) to identify distinct endotypes of atypical diabetes.Methods: This cross-sectional, observational study included RADIANT participants meeting predefined atypical diabetes criteria (n=300). Transcriptomic and metabolomic data were integrated through NEMO (Neighborhood-Based Multi-Omics Clustering), an unsupervised multi-omics clustering algorithm. Cluster separation was evaluated using inter and intra-cluster distances, centroid signatures and Kruskal-Wallis testing with FDR correction.Results: Integrated NEMO clustering yielded 8 subtypes with high reproducibility (mean accuracy 0.82), with 3 highly stable clusters, 3 moderately stable clusters and 2 clusters with overlapping characteristics. These endotypes reflected 3 dominant biological patterns: groups with preserved β-cell function (Clusters 1 and 4), groups indicative of increased insulin resistance with upregulated oxidative, mitochondrial, and ER-stress pathways (Clusters 2, 3, 5, and 6), and groups enriched for MYC, G2M, WNT, and Hedgehog pathways with relatively higher T1D genetic risk scores (Clusters 7 and 8). Phenotypic characteristics and dynamic testing differed among clusters, including BMI (FDR=0.019). T1D polygenic risk scores also varied across clusters (FDR=0.009). Inter- and intra-cluster distances and Prediction Analysis for Microarrays (PAMR) centroids confirmed strong separation. Together, these findings reveal distinct molecular endotypes of atypical diabetes.Conclusion: Unsupervised integration of transcriptomic and metabolomic data identifies phenotypically and molecularly distinct endotypes. Further studies are needed to understand the underlying etiology of these endotypes and develop a more precise classification of atypical diabetes.
J. Faruqi: Research Support; Current; Rhythm Pharmaceuticals, Inc. A. Bareja: None. C. Cheng: None. A. Balasubramanyam: None.
The RADIANT Study is funded by U54 DK118638-07S2 and U54 DK118612 from the National Institute of Diabetes and Digestive and Kidney Diseases. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Source link
Leave a Reply