Deep Immunophenotyping and Clustering Identifies Biomarkers Predictive of Lymphoma in Primary Sjögren Disease

Objective Patients with primary Sjögren disease (pSD) are prone to develop non‐Hodgkin lymphoma (NHL), but relevant biomarkers are lacking. We aimed to determine new biomarkers predictive of NHL in patients with pSD. Methods Two hundred six patients with pSD fulfilling American College of Rheumatology/EULAR 2016 criteria were included and divided into three groups: pSD, lymphoproliferative pSD (Arl‐pSD), and NHL‐pSD. Deep flow cytometry immunophenotyping of B and T cell compartments as well as serum interferon‐α (IFNα) quantification were coupled to clinical, biologic, and histopathologic data analysis. Results We identified CD11c + FcRL5 + tissue‐like memory B cells and IFNγ + TNFα + conventional T cells as significantly associated with NHL in pSD. These clusters showed progressive enrichment in Arl‐pSD and NHL‐pSD as compared to pSD. The combination of these two population abundances discriminates patients with NHL‐pSD with a sensitivity of 78.9% and a specificity of 76.8%, thus overcoming alone the performance of the sum of conventional clinical and biologic markers such as cryoglobulinemia vasculitis, parotid enlargement, adapted clinical EULAR Sjögren's Syndrome Disease Activity Index, rheumatoid factor antibodies, low C4 and elevated serum IFNα levels (68.4% and 78.0%, respectively). CD11c + FcRL5 + tissue‐like memory B cells were associated with occurrence of mucosa‐associated lymphoid tissue (MALT) marginal‐zone NHL, whereas IFNγ + TNFα + conventional T cells were more indicative of non‐MALT B cell NHL. Conclusion We unveil novel biomarkers of NHL in pSD based on an integrative analysis coupling deep immunophenotyping and clinical, biologic, and histopathologic data. Furthermore, these markers allow distinguishing B cell NHL subtypes in pSD.