Abstract
OBJECTIVE: This study aimed to identify immune cell alterations in granulomatosis with polyangiitis patients in remission (rGPA) that may facilitate diagnosis and prediction of relapse. METHODS: Circulating immune cells were phenotypically characterized by high-dimensional CyTOF in 59 rGPA patients and 31 healthy controls (HCs). These data together with inducible cytokine expression assays and Machine Learning (ML) methods were used to identify immunophenotypic profiles distinguishing rGPA patients from HCs and patients with higher relapse frequencies. RESULTS: rGPA patients exhibited multiple blood cell immunophenotypic features distinct from HCs, including lymphocytopenia, a shift toward exhausted effector T cells and increased B and innate immune cell activation. Using ML methods, we identified a combination of cell features (γδ T cell depletion, monocyte and CD177(+) neutrophil expansion, B cell depletion) distinguishing rGPA patients from HCs and cytokine expression profiles among patients (increased IL-8 in monocytes, decreased IL-10 in monocytes and cDC2 cells) associated with relapse frequency. Two ML-based risk scores were developed and respectively shown to accurately discriminate rGPA cases from HCs and rGPA patients with more frequent disease relapse. CONCLUSIONS: Our findings reveal distinct patterns of immune dysregulation in rGPA patients and demonstrate potential for ML methods to facilitate disease diagnosis and outcome prediction based on immunophenotypic data.