SAT0051 REMISSION IN RHEUMATOID ARTHRITIS PATIENTS: A CLUSTER ANALYSIS TO IDENTIFY AND CHARACTERIZE SUBPOPULATIONS OF PATIENTS. (2nd June 2020)

• Record Type: Journal Article
• Authors: Madrid García, A.; Freites Nuñez, D.; Font, J.; Hernandez, I.; León, L.; Colomer, J. I.; González-Álvaro, I.; Fernandez, B.; Jover, J. A.; Abasolo, L.; Rodriguez Rodriguez, L.
• Is Part Of: Annals of the rheumatic diseases. Volume 79(2020)Supplement 1
• Page Start: 956
• Page End: 957
• Publication Date: 2020-06-02
• Subjects: Rheumatism -- Periodicals; 616.723005
• DOI: 10.1136/annrheumdis-2020-eular.5750
• Languages: English
• ISSNs: 0003-4967

Abstract

Abstract : Background: Disease Activity Score (DAS) is a continuous measure of Rheumatoid Arthritis (RA) activity, used in clinical practice for monitoring disease progression and for documenting treatment response. According to EULAR, the clinical desired target is to achieve a remission state (or failing that, low disease activity). However, the population of RA patients in this state could be heterogenous. Objectives: To characterize the level of heterogeneity of RA patients in remission by identifying clusters based on the DAS28 components; and to describe inter and intra-class cluster demographic and clinical characteristics Methods: Patients from Hospital Clínico San Carlos cohort, stored in a departmental electronic health record from January 1 st, 2000 to December 30 th, 2018, diagnose with RA according ACR 1986/ 2010 criteria were eligible for this study. Only observations with a DAS28 Erythrocyte Sedimentation Rate (ESR) value < 2.6 value were considered. ESR, patient's Global Health (GH), and tender and swollen joints were used for calculating the clusters. Different aggregation levels for joints were studied as well as the input variable types. Isolated joints, joints grouped by the type of affectation (swollen or tender) or anatomic location or laterality aggregation levels were considered. Variables expressed as present or absent (i.e. dichotomous), continuous (count of joins) and categorical (type of joints) were also studied. Gower's distance, used for dealing with variables of different type, was employed to calculate the distance matrix. The number of suitable clusters was chosen from two to seven clusters based on the width value of a Silhouette analyses. Finally, Partitioning Around Medoids (PAM) was used as the clustering algorithm. Differences among clusters regarding demographic and clinical characteristics were analyzed using t-student chi2 test. Results: 812 patients with 1, 431 observations were analyzed in this study. The joint aggregation level which showed a highest Silhouette width value (0.708) was the anatomic one. In this aggregation level, five dichotomous variables (presence of tenderness and/or swelling in right and/or left shoulder, elbow, wrist, knee and hand (including both metacarpophalangeal and proximal interphalangeal joints) and two continuous variables (ESR and GH) were used. Two clusters were found: the cluster A) with 1, 305 observations and 742 patients and the B) with 126 observations and 115 patients. Cluster b) had a statistically significant higher DAS28-ESR value (higher number of tender and swollen joints, and higher GH, but lower ESR), longer follow-up time (6.5 vs. 4.7 years), higher VAS-pain score (10 vs. 2), and higher HAQ score (0.25 vs. 0.12). In addition, the proportion of patients treated with oral corticosteroids (63% s. 50%) and biological therapy (29% vs. 12%) was higher. Conclusion: We have identified two clinically distinct populations of RA patients in remission according to DAS28-ESR <2.6. Each subgroup could be associated with different outcomes during follow-up, such as radiographic progression or risk of relapse. Disclosure of Interests: Alfredo Madrid García: None declared, Dalifer Freites Nuñez: None declared, Judit Font: None declared, Isabel Hernandez: None declared, Leticia León: None declared, Jose Ignacio Colomer: None declared, Isidoro González-Álvaro Grant/research support from: Roche Laboratories, Consultant of: Lilly, Sanofi, Paid instructor for: Lilly, Speakers bureau: Abbvie, MSD, Roche, Lilly, Benjamin Fernandez: None declared, J Angel Jover: None declared, Lydia Abasolo: None declared, Luis Rodriguez Rodriguez: None declared