1025-OR: Immune Repertoire Profiling for Accurate Classification of Latent Autoimmune Diabetes in Adults

Introduction and Objective: Latent Autoimmune Diabetes in Adults (LADA) shares features of both type 1 (T1D) and type 2 diabetes (T2D), leading to frequent misdiagnosis with current tools. This study aimed to determine whether T-cell and B-cell receptor (TCR/BCR) repertoire profiling could serve as a clinically actionable biomarker to accurately distinguish LADA Methods: We performed deep TCR and BCR sequencing on peripheral blood mononuclear cells (PBMCs) samples from patients with LADA (n = 20), T1D (n = 20), T2D (n = 15), and healthy donors (HD, n = 20), with an independent validation cohort comprising 15 LADA and 15 T1D patients. Associations between TCR/BCR repertoires and clinical/cytokine profiles were analyzed. Machine learning models were built using repertoire features to discriminate LADA from T1D. Results: LADA patients exhibited a distinct immune repertoire signature, characterized by significantly higher TCRα/β and BCR (particularly IgM and IgG) repertoire diversity, and the lowest clonal dominance compared to T1D, T2D, and HD (all P<0.0001). This hyper-diverse, less clonal immune state in LADA was associated with a unique cytokine profile, featuring the highest levels of IL-2 and IFN-γ and the lowest IL-6 (all P<0.0001). Significant correlations were found between repertoire features and key metabolic biomarkers, autoantibody titters, and cytokines. Machine learning models integrating TCR/BCR repertoire features achieved superior performance (AUC = 0.889) in distinguishing LADA from T1D in both training and validation sets, outperforming models based solely on conventional autoantibodies (AUC = 0.667) or clinical indicators (AUC = 0.680). Conclusion: TCR/BCR repertoire profiling is a highly accurate and novel diagnostic tool that can effectively differentiate LADA from T1D, offering a direct path to more precise patient classification and personalized management in clinical practice. Disclosure X. Lyu: None. R. Qi: None. L. Yao: None. L. You: None. J. Yan: None. C. Hu: None. W. Jia: None. Funding National Natural Science Foundation of China (82570944) and Shanghai 2022 High level Expert Support Special Foundation (22YS1400500)