W108 - Blood TCR Signatures Classify Autoimmune Diseases and Predict Response to Low-dose IL-2

Encarnita Mariotti-Ferrandiz; Pierre Barennes; Paul Stys; Roberta Lorenzon; Nicolas Coatnoan; Kenz Le Gouge; Hélène Vantomme; Michelle Rosenzwagj; Agnes Hartemann; jeremie Sellam; Francis Berenbaum; David Klatzmann

Abstract Text: Autoimmune diseases (ADs) are chronic and debilitating diseases representing an important societal burden. They call for disease-specific and curative treatments as well as for better diagnostic and prognostic markers. ADs result from patient’s T and B lymphocyte attacking their own tissues. While autoantibodies are used for AD diagnosis, T-cell are not considered for diagnosis in spite of their critical role in AD onset and maintenance. To tackle this major gap, we analyzed the T-Cell Receptor (TCR) repertoire by next-generation sequencing in various AD contexts. We sorted CD4 T effector cells (Teff) and CD4 T regulatory T-cells (Treg) from the blood and analyzed their TCRs using an innovative machine learning approach based on sparse least square discriminant analysis.
When applied to type 1 diabetes (T1D) and rheumatoid arthritis (RA) patients, we identified Teff and Treg TCR signatures that i) classify these patients versus healthy volunteers with > 90% accuracy and ii) properly cluster T1D and RA patients separately. These signatures were shown to contain TCRs that were previously shown to relate to these conditions and were validated on external datasets.
We also applied our analyses to sequential samples from lupus patients treated with low-dose IL-2. We identified a Treg TCR signature that could predict clinical response to treatment.
Altogether, these results show that peripheral blood TCR repertoires contain relevant disease-specific information that could serve as biomarker to improve AD diagnosis, prognosis and patient care.