Cluster analysis of clinical manifestations assigns systemic lupus erythematosus-phenotype subgroups: A multicentre study on 440 patients

Objective

Systemic lupus erythematous (SLE) is a heterogenous disease characterised by a large panel of autoantibodies and a wide spectrum of clinical signs and symptoms that engender different outcomes. We aimed to identify distinct, homogeneous SLE patients’ phenotypes.

Methods

This retrospective study enrolled SLE patients meeting the Systemic Lupus International Collaborating Clinics (SLICC) classification criteria, enrolled in the French multicentre “APS (antiphospholipid syndrome) and SLE” Registry. Based on 29 variables selected to cover a broad range of clinical and laboratory (excluding autoantibodies) SLE manifestations, unsupervised multiple correspondence analysis followed by hierarchical ascendent-clustering analysis assigned different phenotypes.

Results

We included 440 patients, mostly women (94.3%). Median age at SLE diagnosis was 24 (IQR 19–32) years. Cluster analysis yielded three distinct subgroups based on cumulative clinical manifestations, not autoantibody pattern. Cluster 1 (n = 91) comprised mostly Caucasian patients, with APS-associated clinical and biological manifestations, e.g., livedo, seizure, thrombocytopaenia and haemolytic anaemia. Cluster 2 (n = 221), the largest, included patients with mild clinical manifestations, mainly articular, more frequently associated with Sjögren's syndrome and with less frequent autoantibody-positivity. Cluster 3 (n = 128) consisted of patients with the largest panel of SLE-specific clinical manifestations (cutaneous, articular, proliferative nephritis, pleural, cardiac and haematological), the most frequent autoantibody-positivity, low complement levels, and more often of Asian and sub-Saharan African origin.

Conclusion

This unsupervised clustering method distinguished three distinct SLE patient subgroups, highlighting SLE heterogeneity.