1129: Integrated High-throughput Proteomics and Machine Learning Analysis in Systemic Lupus Erythematosus Patients Identify Distinctive Clinical Profiles and Novel Biomarkers Related to Cardiovascular Risk and Lupus Nephropathy

Chary Lopez-Pedrera¹, Tomás Cerdó², María Ángeles Aguirre-Zamorano³, Laura Muñoz-Barrera¹, Ismael Sánchez-Pareja², Pilar Font¹, María Carmen Ábalos-Aguilera⁴, Pedro Ortiz-Buitrago⁴, Nuria Barbarroja¹, Eduardo Collantes¹, Rafaela Ortega-Castro¹ and Carlos Pérez-Sánchez¹, ¹IMIBIC/University of Cordoba/Reina Sofia Hospital, Cordoba, Spain, ²IMIBIC/University of Cordoba/Reina Sofia Hospital, Córdoba, Spain, ³Reina Sofía University Hospital/ Rheumatology Department, Córdoba, Spain, ⁴Imibic/ University of Córdoba/ Reina Sofía Hospital, Córdoba, Spain

Background/Purpose: Systemic lupus erythematosus (SLE) is a remarkably heterogeneous autoimmune disease. At present, our knowledge of serum protein patterns related to cardiovascular (CV) risk and renal involvement is still limited.To characterize clinical phenotypes in SLE patients through serological analysis of proteomic profiles.

Methods: Proximity extension immunoassay (PEA, Olink) was used to assess the serum levels of one hundred and eighty-four inflammation and organ damage-related proteins in patients with SLE (n = 141) and age-matched healthy donors (HD) (n = 28). In parallel, an extensive clinical and analytical profile of recruited subjects was performed. To evaluate the contribution of molecular profiles to disease features, unsupervised machine learning clustering analyses were developed. Gene ontology enrichment analysis were also carried out to interrogate the biological meaning associated with the molecular signatures identified.

Results: Several circulating proteins related to inflammation and organ damage were coordinately altered in the serum of SLE patients in relation to HD. Unsupervised clustering analyses differentiated 2 patients clusters presenting different proteomic profiles. Clinically, although no differences were found in terms of age, gender, disease duration, or treatments, patients belonging to cluster 1 were characterized by higher status of disease activity (SLEDAI over 5,3) and prevalence or positivity for anti-ENA and anti-dsDNA antibodies than patients belonging to cluster 2. These patients showed a preponderance of lupus nephritis (LN) and proteinuria. Besides, this cluster comprised SLE patients with higher CV risk, revealed by an elevated incidence of dyslipidemia, obesity and hypertension. Sixty-seven proteins were found deregulated between clusters, including inflammatory mediators (cytokines, chemokines and regulatory proteins of leukocyte activity) and numerous proteins related to both, renal disease and increased CV risk. Of note, fifty proteins were found significantly altered between patients with LN vs patients without LN. Logistic regression analyses identified several proteins which levels distinguished lupus nephritis patients with high accuracy, including several proteins involved in renal damage, not previously reported in the serum of SLE patients.

Conclusion: 1) This highly sensitive proteomic analysis in the serum of SLE identified molecular patterns distinguishing patients with high disease activity and active LN, and including several novel candidate proteins, whose exact role and suitability as biomarkers in SLE deserve further investigation. 2) Combination of novel and traditional disease-specific biomarkers may improve diagnosis and management of SLE.

Supported by ISCIII (PI21/005991 and RICOR-RD21/0002/0033) co-financed by FEDER

Disclosures: C. Lopez-Pedrera, None; T. Cerdó, None; M. Aguirre-Zamorano, None; L. Muñoz-Barrera, None; I. Sánchez-Pareja, None; P. Font, None; M. Ábalos-Aguilera, None; P. Ortiz-Buitrago, None; N. Barbarroja, None; E. Collantes, None; R. Ortega-Castro, None; C. Pérez-Sánchez, None.