Abstract
Teplizumab, a humanized anti-CD3 mAb, represents a breakthrough in autoimmune type 1 diabetes (T1D) treatment, by delaying clinical onset in stage 2 and slowing progression in early stage 3 of the disease. However, therapeutic responses are heterogeneous. To better understand this variability, we applied single-cell transcriptomics to paired peripheral blood and pancreas samples from anti-mouse CD3-treated nonobese diabetic (NOD) mice and identified distinct gene signatures associated with the therapy outcome, with consistent patterns across compartments. Success-associated signatures were enriched in NK or CD8+ T cells and other immune cell types, whereas resistance signatures were predominantly expressed by neutrophils. The immune cell communities underlying these response signatures were confirmed in human whole blood sequencing data from the Autoimmunity-blocking Antibody for Tolerance (AbATE) study at 6 months, which assessed teplizumab therapy in individuals with stage 3 T1D. Furthermore, baseline expression profiling in the human TrialNet Anti-CD3 Prevention (TN10) (stage 2) and AbATE (stage 3) cohorts identified immune signatures predictive of therapy response, T cell-enriched signatures in responders, and neutrophil-enriched signatures in nonresponders, highlighting the roles of both adaptive and innate immunity in determining teplizumab treatment outcomes. Using an elastic net logistic regression model, we developed a 26-gene blood-based signature predicting the response to teplizumab (AUC = 0.97). These findings demonstrate the predictive potential of immune gene signatures and the value of transcriptomics profiling in guiding individualized treatment strategies with teplizumab in individuals with T1D.