SLE CLASSIFICATION CRITERIA ITEM RELATIONSHIPS: IMPLICATIONS ON SLE AS A DISEASE ENTITY

PV216 / #355 Poster Topic: AS23 - SLE-Diagnosis, Manifestations, & Outcomes Background/Purpose To investigate for clusters and associations between the European Alliance of Associations for Rheumatology (EULAR)/American College for Rheumatology (ACR) classification criteria items. Methods Multiple Correspondence Analysis (MCA) was performed on the 10 classification criteria domains, using R (version 4.3.2) and FactoMineR (version 2.11) package in an international cohort of 1,197 SLE patients. Consensus clustering was performed using the ConsensusClusterPlus package (Version 1.66.0). Associations between individual criteria items were analyzed in the full 23x23 table (Graph Pad prism), and, by Bonferroni’s correction, statistical significance was defined as p<9.45e-5. Results SLE patients fulfilled a minimum of zero (n=1) and a maximum of 15 of the 23 items (median 6) within 0 to 9 of the 10 domains (median 4). More than 2/3 of the patients (presently and/or historically) had criteria items within the domain SLE-specific antibodies (79.8%), the mucocutaneous (72.0%) and musculoskeletal (71.9%) domains, and complements (71.6%). In MCA, constitutional, musculoskeletal, and serosal domains clustered together, separating from the renal domain, complements, and SLE-specific antibodies (Figure 1A). However, the distribution of individual patients based on the first 2 dimensions (Figure 1B) was not reflective of a relevant separation of the cohort within those 2 dimensions. Consensus clustering, set at 10 clusters according to the cumulative distribution function (CDF, Figure 2A and 2B), revealed clusters with various combinations of domains (Figure 2C, proportion of patients with items in the respective domain indicated), but without a clear, mutually exclusive pattern. Among the 529 pairwise comparisons of items, 27 pairs showed a positive association, with r-values ranging from 0.11 to 0.58, and 4 pairs a negative association (r-values from -0.13 to -0.16). All r-values of 0.3 or higher were within an organ domain. An r-value of ≥2.0 was found for anti-dsDNA, proteinuria, and proliferative (class III or IV) nephritis, each with low complements, in addition to associations between domains. Figure 1. MCA. Figure 2. Consensus Clustering. Conclusions We found distributions more compatible with chance associations of domains and items than with true SLE subsets. The associations between items within domains support the decision to use domains in the EULAR/ACR classification criteria structure.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.