POS0051 CLINICAL AND IMMUNOLOGICAL ENDOTYPES IN SJÖGREN'S DISEASE: DATA FROM BRAZILIAN REGISTRY OF SJÖGREN'S DISEASE (BRAS) USING MCA AND HIERARCHICAL CLUSTERING

Résumé fourni par la source

Background: Sjögren's disease (SjD) is mainly characterized by the involvement of the exocrine glands, but around 40% of patients have systemic involvement. There is an association between systemic involvement, B-cell hyperreactivity, and the development of lymphoma. Recent studies have shown the existence of distinct subgroups characterized by the presence or absence of symptoms, clinical manifestations, and biological findings. Objectives: Our study aimed to characterize patients with SDj from the Brazilian Sjögren's Disease Registry (BRAS) into subgroups using a cluster analysis based on their clinical and serological findings. Methods: This study utilized data from a multicenter database hosted on the REDCap platform from the Brazilian Sjögren's Disease Registry (BRAS). A total of 276 patients fulfilling the 2002 and 2016 criteria for SDj were included in the analysis. The dataset was carefully curated to ensure completeness, with missing data addressed to create a robust final table suitable for statistical modeling. To prepare the data for Multiple Correspondence Analysis (MCA), continuous variables were categorized, making them compatible with the method's requirements. Variables were selected based on their clinical relevance to minimize noise from less important features. A total of 12 variables were included, covering demographic, clinical, and laboratory data such as age, ESSPRI, ESSDAI, cryoglobulin levels, and anti-Ro positivity. MCA was conducted to reduce the dimensionality of the dataset, identifying key patterns while retaining the most relevant information. Following dimensionality reduction, hierarchical clustering using the Ward.D2 method was applied to group patients into clinically interpretable clusters. The optimal number of clusters was determined using dendrograms, WSS plots, silhouette scores, and gap statistics. This process enabled a clear visualization of subgroup structures within the patient population. All statistical analyses were performed using R (version 4.3.0). The FactoMineR library was central to conducting MCA and clustering, while additional R packages supported data visualization and validation of cluster assignments. Results: The MCA reduced the dimensionality of the dataset, with the first components explaining a significant portion of the variance. Four distinct clusters were identified, each representing unique clinical and immunological profiles. The first cluster included patients with prominent symptoms but minimal systemic or serological abnormalities, suggesting a symptomatic presentation with limited immune involvement. The second cluster consisted of patients with mild symptoms and low clinical and laboratory activity, representing a less severe disease form. In contrast, the third cluster was characterized by high systemic and serological activity, with pronounced clinical manifestations and elevated laboratory markers indicative of a systemic disease endotype. The fourth cluster exhibited the most extensive serological abnormalities, reflecting a strong B-cell response. Interestingly, there was a decrease in ESSDAI over time in clusters 1, 2, and 3, while there was an increase in ESSDAI in cluster 4. The highest levels of fatigue were in cluster 1 (high symptom, low systemic) but also cluster 3 (high systemic). The only systemic domain associated with cluster 1 (high symptom, low systemic) was articular domain. Considering the short follow-up time and the small sample size, no correlation was found with death or lymphoma. Conclusion: These results underscore the heterogeneity of SjD, revealing distinct endotypes that could guide future research and therapeutic approaches. Figure 1 Figure 2Categorical and Numerical Variable Behavior by Cluster REFERENCES: NIL . Acknowledgements: NIL . Disclosure of Interests: None declared . © The Authors 2025. This abstract is an open access article published in Annals of Rheumatic Diseases under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Neither EULAR nor the publisher make any representation as to the accuracy of the content. The authors are solely responsible for the content in their abstract including accuracy of the facts, statements, results, conclusion, citing resources etc.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.