Background/Purpose: Systemic lupus erythematosus (SLE) is a heterogeneous autoimmune disease characterized by diverse clinical presentation, posing significant challenges in diagnosis and management. Reflecting this clinical complexity, multiple sets of classification criteria have been developed for SLE. Genetic factors contribute significantly to SLE susceptibility, with genome-wide association studies (GWAS) identifying over 130 distinct genetic loci linked to increased disease risk. Importantly, polygenic risk scores (PRSs) derived from these loci have been associated with specific clinical manifestations, including cardiovascular and renal involvement, indicating that genetic risk may drive distinct phenotypic expressions within SLE. Here, we developed an SLE PRS and evaluated its associations with electronic health record (EHR)-derived clinical phenotypes and laboratory parameters available in the Electronic Medical Records and Genomics (eMERGE) Network biobank. We further assessed the relationships between the PRS and individual Systemic Lupus International Collaborating Clinics (SLICC) classification criteria.

Methods: We modeled polygenic risk for lupus using PRS-CS on summary statistics from a large genome-wide association study consisting of 6,748 SLE cases and 11,516 controls of European ancestry and derived PRSs in European ancestry eMERGE participants. We performed a phenome-wide association study between the lupus PRS and 1,706 phecodes in Northwestern University eMERGE participants (n=1,061), adjusting for age, sex, number of encounters, and the top 3 principal components of ancestry. We then calculated correlations between the lupus PRS and the prevalences of individual SLICC criteria across all eMERGE sites (n=32,329), adjusting for the number of encounters.

Results: The PRS predicted SLE with a receiver operating characteristic area under the curve of 0.724. Within the Northwestern cohort, the PRS was significantly associated with “Lupus (localized and systemic)” (P=1.6×10-9) and “systemic lupus erythematosus” (P=3.0×10-9). Nominally associated phecodes (P < 5.0x10-3) included “cardiac complications” (P=2.8x10-3) and “impaired renal function” (P=4.6x10-3). PheWAS associations were significantly enriched for circulatory system (P=9.7x10-5) and hematopoietic (P=3.2x10-2) traits. Across eMERGE, the SLE PRS was significantly associated with 5/6 immunological and 3/11 clinical SLICC criteria (Acute cutaneous lupus, thrombocytopenia, and leukopenia). Of the SLICC criteria, the PRS was most strongly associated with low complement levels (P=7.4x10-31), with nearly twice the rate of lupus patients with low complement levels in the highest PRS decile than those in the bottom PRS decile (55% vs. 30%, respectively).

Conclusion: Our findings indicate that different clinical and immunological criteria are more or less associated with genetic SLE risk. Future studies should consider incorporating genetic risk for developing more comprehensive and accurate prediction models of SLE to inform strategies for clinical prevention, intervention, and management.

« Back to ACR Convergence 2025

ACR Meeting Abstracts - https://acrabstracts.org/abstract/polygenic-risk-of-lupus-is-differentially-associated-with-individual-ehr-derived-classification-criteria/