Three clinical and immunological phenotypes of systemic lupus erythematosus onset identified through clustering analysis: data from the RELES registry

Abstract

Abstract Background Systemic lupus erythematosus (SLE) shows heterogeneous clinical expression at onset, complicating early characterisation and risk assessment. We aimed to identify SLE phenotypes emerging after diagnosis and assess their clinical relevance. Methods We analysed newly diagnosed SLE patients from the Spanish multicentre RELES inception cohort. Eighteen binary clinical and immunological variables based on the 2019 EULAR/ACR classification criteria, cumulatively recorded within two years of diagnosis, were included. Multiple correspondence analysis followed by unsupervised hierarchical clustering on principal components with k-means consolidation was performed. Internal validation metrics assessed cluster stability and separation. Baseline characteristics, treatments and early clinical burden were compared across clusters. Results A total of 795 patients were included (87.7% women; mean age at diagnosis 42 ± 16 years). Three clinically interpretable phenotypes were identified. Cluster 1 ( n = 168, 21.1%) represented a renal-dominant phenotype, characterised by younger age, frequent lupus nephritis, complement consumption and greater exposure to intensive immunosuppression, with the highest rates of infections and cumulative damage. Cluster 2 ( n = 175, 22.0%) showed a haematologic-serosal phenotype, with cytopenias, serositis, and frequent antiphospholipid and anti-dsDNA positivity, also associated with substantial early damage. Cluster 3 ( n = 452, 56.9%) represented a mild cutaneous-articular phenotype, with lower serological activity and the lowest burden of early complications. Internal validation supported a stable three-cluster solution, with moderate separation reflecting the continuum of early SLE. Conclusion Three clinically meaningful SLE phenotypes can be identified within two years of diagnosis using routinely collected variables. This approach may provide a practical framework for early disease characterisation and risk-adapted management in SLE.