Document resource
Objectives Using unsupervised clustering analysing, we aimed to identify autoantibody-defined subgroups at diagnosis in a Systemic Lupus Erythematosus (SLE) inception cohort and assess their associations with clinical manifestations and outcomes.Methods We used the Nor-SLE population-based inception cohort of chart-confirmed SLE cases in South-East Norway 2000–2017 fulfilling the 2019 EULAR/ACR (EA) SLE classification criteria. Patients were clustered by eight autoantibodies at diagnosis (ELISA or immunoprecipitation (IP); positivity by either. Anti-dsDNA was positive by ELISA and/or IP; anti-β2GPI and anticardiolipin were positive by IgM and/or IgG. Using logistic regression adjusted for sex and age, we tested associations between clusters, EA criteria, Raynaud phenomenon, antiphospholipid syndrome and arterial/venous thromboembolism assessed at diagnosis and last visit (2017). Extending follow-up to 2022, we assessed mortality by antibody-cluster with standardized mortality ratios (SMR) using 15 population controls per patient matched by age, sex and ancestry.Results Of 700 patients, 358 (51%) had a complete antibody profile at diagnosis, 26 (4%) were negative for all eight antibodies, leaving 332 (47%) for clustering. Four clusters showed the best separation of the patients based on the average Silhouette scores ( figure 1A). Patients who were negative for all eight antibodies at diagnosis were included as cluster five. Cluster two had the highest proportions of patients of non-European ancestry (table 1).At diagnosis, patients in cluster two had a higher risk of pericarditis (OR 2.79, 95%CI 1.31-5.71) and Raynaud phenomenon (OR 2.19, 95%CI 1.27-3.79) (figure 1B) versus the other clusters (figure 1B). At last visit, after median six years follow-up (IQR 3-0), patients in cluster four showed increased risk of epilepsy (OR 3.99, 95% CI 1.21-14.23) (figure 1C).Over median 11 years follow-up (IQR 7–15), 26/358 (7%) SLE patients died versus 329/5370 (6%) controls. Seven deaths occurred in cluster two and nine in cluster four. In the remaining clusters the number of deaths were too low to report. The overall SMR for SLE was 1.6 (95% CI 1.1–2.5), higher in cluster two at 3.7 (95% CI 1.4–8.7) and lower in cluster four at 1.3 (95% CI 0.6–2.6).Abstract S10:01 Table 1Demographic, clinical and immunological parameters by auto-antibody clusters at time of systemic lupus erythematosus diagnosisAbstract S10:01 Figure 1Auto-antibody clusters at systemic lupus erythematosus diagnosis and their associations with clinical manifestations and outcomes at diagnosis and last follow-upConclusions Five distinct autoantibody clusters at the time of SLE diagnosis were associated with clinical sub-phenotypes and mortality in SLE.