SUN-565 High-dimensional Immunophenotyping of Immune Checkpoint Inhibitor-Induced Type 1 Diabetes

Abstract Disclosure: Z. Quandt: None. S. Yellamilli: None. E. Flynn: None. V. Johri: None. B. Davidson: None. J. Tsui: None. A. Combes: None. G. Fragiadakis: None. A. Young: None. M. Anderson: None. Introduction: Immune checkpoint inhibitor (ICI) type 1 diabetes (T1DM) is a rare but morbid immune related adverse event (irAE) that can occur in patients with cancer following immunotherapy. ICI-T1DM almost exclusively results from immune checkpoint inhibitors targeting the PD-1/PD-L1 pathway rather than CTLA-4. While some shared features of spontaneous T1DM, such as HLA type and presence of islet cell autoantibodies, are apparent in some, but not all, ICI-T1DM; however, little is known about the immunophenotypic characteristics that predispose patients with cancer to risk of ICI-T1DM as compared to healthy individuals and spontaneous T1DM. Methods: We collected peripheral blood mononuclear cells from 69 patients, 10 with recent onset ICI-T1DM, 17 with recent onset T1DM and 42 healthy controls and performed CITEseq. Individual libraries were processed using a standard single-cell analysis pipeline for RNA and TCR-sequencing alongside protein integration for antibody-derived tags (CellRanger for alignment, followed by Seurat for quality control and initial processing). Multiplexed samples were demultiplexed using freemuxlet/demuxlet, and intrasample doublets were removed using doubletFinder. Individual libraries were integrated using Harmony for RNA data and dsbNormalize with reciprocal principal components (RPCA) for protein (ADT) data. A combined embedding was generated using Weighted Nearest Neighbors, followed by UMAP visualization and clustering. Differential expression analysis was performed using EdgeR with subsequent pathway analysis using Gene Ontology (GO) terms and Reactome pathways. Results: Differential gene expression analysis revealed enhanced MHC-II pathway activity in ICI-T1DM patients compared to healthy individuals. Sub-clustering of T cells identified five distinct clusters with differential abundance between ICI-T1DM, T1DM, and healthy controls. One cluster showed significant enrichment for a T follicular helper (Tfh) cell signature previously implicated in ICI-T1DM in the NOD mouse model (Huang et al. 2024). The proportion of these cells was significantly higher in ICI-T1DM patients compared to both individuals with recent onset spontaneous T1DM and healthy individuals (FDR adjusted p=0.02). Conclusions: These findings suggest a potential activation state in the T cell compartment by upregulation of HLA genes and the expansion of Tfh cells that correlate with the development of ICI-T1DM in patients with cancer. We will next validate these findings in a secondary cohort of ICI-treated patients with cancer that do not have ICI-T1DM. Presentation: Sunday, July 13, 2025