DIFFERENT PHENOTYPES OF SEVERE FLARES IN PATIENTS WITH SYSTEMIC LUPUS ERYTHEMATOSUS (SLE): RESULTS OF A CLUSTERING ANALYSIS IN A MONOCENTRIC COHORT

Background/Purpose To identify and compare different phenotypes of severe flares in a monocentric cohort of SLE patients. Methods This is a retrospective study of prospectively collected data from a monocentric cohort of adult SLE patients (2019 EULAR-ACR classification criteria), hospitalized in the last 5 years, due to a severe flare (SELENA-SLEDAI flare index definition). Patients with concomitant infections, oncologic and onco-hematologic conditions were excluded. Hospitalization was defined as baseline (t0). At t0 demographics, clinical, laboratory and treatment data were collected. Disease activity was assessed with SLEDAI-2K and BILAG-2004. Disease outcomes (Lupus Low Disease Activity State (LLDAS) and DORIS remission) and treatment were evaluated at 3, 6, 12-months (t3, t6, t12) after the flare. Organ damage (SLICC Damage Index (SLICC-DI)) was assessed at baseline and t12. A clustering analysis was performed on SLE flares, with a hierarchical method. Post hoc elaborations (1-way analysis of variance with Bonferroni test for quantitative variables, and Chi-square test for qualitative variables) were performed to estimate any statistically significant differences between the clusters. Results 122 severe flares in 110 patients (female 83%, Caucasian 89%) were included. 3 clusters were identified, composed of 40, 34 and 48 flares respectively. Cluster 1 included flares that occurred in younger patients (mean age 38.2±12 vs 46.3±13.9 and 43.2±11.1 in clusters 2 and 3 respectively; p=0.007) with a shorter disease duration (9.1±6 years vs 17.5±11.5 and 15±8.7; p=0.0001), characterized by a higher frequency of BILAG A manifestations in the constitutional, cardiopulmonary and musculoskeletal domains (p= 0.0001). These flares presented hyperinflammatory stigmata (higher C-reactive protein, more severe lymphopenia, a tendency for higher ferritin values) and a richer autoantibody profile (anti-dsDNA, anti-Smith, anti-nucleosome, anti-hystone), compared to the other clusters. Cluster 2 included less severe flares with more BILAG B scores (59% vs 28% and 19%; p=0.0001) and mainly joint and skin manifestations. Cluster 3 was characterized by a clear predominance of renal flares (96%) (p=0.0001). 85% of flares in each cluster required adding/changing the immunsuppressant, mainly Mycophenolate in clusters 1 and 3 (30% and 47% vs 9%; p=0.001). Glucocorticoid pulses were less frequently used in cluster 2, accordingly to a milder flare phenotype (15% vs 50% and 83%; p= 0.0001). Belimumab was added in 25% and 30% of flares in cluster 1 and 2 respectively, only in 8.5% in cluster 3 (p=0.02), as the majority of flares occurred before the approval of Belimumab for lupus nephritis. At t6 and t12, flares in clusters 1 and 3 presented a significantly higher cumulative glucocorticoid dose, compared to cluster 2. At the different timepoints, cluster 1 and 3 presented a comparable and quite low percentage of patients that achieved LLDAS and remission (Table 1). No differences emerged among the clusters for SLICC-DI at t12. Conclusions Different phenotypes of severe SLE flares exist. We identified a “hyperinflammatory” phenotype presenting with fever, arthritis and serositis, deserving similar aggressive therapeutic strategies as renal flares and burdened by a comparable proportion of unsatisfying response to treatment.