A LONGITUDINAL ANALYSIS OF CHANGE IN LUPUS DISEASE ACTIVITY PATTERN IN HOPKINS LUPUS COHORT USING A MULTISTATE MARKOV MODEL APPROACH

Background: Systemic Lupus Erythematosus (SLE) is a multi-systemic inflammatory disease with extreme variability of its activity over time. We have described three main patterns: long quiescence (LQ), chronic activity (CA) and relapsing-remitting (RR). Patients may move through a series of these activity states over time. To capture these complex fluctuations, we used a Multistate Markov model to evaluate SLE disease activity patterns over time in a longitudinal lupus cohort. Objectives: To describe burden of disease activity patterns over time among patients with SLE. To estimate the probability of transition between disease activity patterns and survival at 1, 2 and 5 years after SLE diagnosis. To identify predictors of change in disease activity patterns. Methods: SLE patients were followed up quarterly in Hopkins Lupus Cohort for 1–28 years. Medication, disease activity (by Physician Global Assessment (PGA) and SLE Disease Activity Index (SLEDAI)), C3, C4, anti-dsDNA, antiphospholipid antibodies and urine protein/creatinine ratio were recorded at each visit. For each patient, visits were divided into 1-year blocks. Any 1-year block with only one visit or patients followed for only 1 year were excluded. SLE disease activity patterns were defined using PGA and SLEDAI. LQ: SLEDAI/PGA=0 for all visits within 1-year blocks; RR: periods of disease activity (SLEDAI/PGA>0) interspersed with periods of disease inactivity (SLEDAI/PGA=0) within 1-year block; CA: SLEDAI/PGA>0 for all visits within 1-year block. Multistate Markov models were used to provide estimates of relative transition rates and identify predictors of change in disease activity patterns. Results: 1735 SLE patients were included in this analysis. 92.3% were females, mean ± SD age 32.8±13.1 years at diagnosis. 52.9% were Caucasian and 40.3% African American, the remainders were mostly Asian. 127 patients died during follow up. Likelihood of deterioration (LQ to RR; RR to CA; LQ to CA) was greater than improvement (RR to LQ; CA to RR; CA to LQ). At year 1, 63.6% remained in LQ, while at year 2 and year 5, 58.2% and 84.1% had deteriorated from LQ pattern to either RR or CA pattern. From the RR pattern, the estimated probability for deterioration (to CA pattern) was higher than improvement (to LQ pattern), regardless of years since SLE diagnosis. Multivariate analysis (see table 1) identified that race/ethnicity (African American), low C3, low C4, ESR >20, taking prednisone, plaquenil or anti-hypertension drug were predictors for deterioration. Plaquenil improved survival. Table 1. Hazard ratio and 95% confidence interval of predictors of transitions RR → CARR → LQRR → Death African American vs. Caucasian1.81 (1.59,2.06)*0.72 (0.58,0.89)1.57 (0.93,2.64) Urine protein creatine ratio1.11 (0.96,1.28)0.48 (0.34,0.68)*1.79 (1.11,2.89) Low C31.75 (1.51,2.03)*0.38 (0.24,0.59)*1.46 (0.82,2.59) Low C41.56 (1.32,1.83)*0.2 (0.1,0.4)*1.69 (0.88,3.26) ESR.c1.28 (1.13,1.45)*0.68 (0.56,0.81)*2.4 (1.34,4.29) Prednisone1.06 (0.94,1.19)0.65 (0.54,0.78)*3.83 (1.99,7.38)* Plaquenil1.28 (1.12,1.45)*0.98 (0.81,1.18)0.37 (0.23,0.59)* Anti hypertension drug1.08 (0.96,1.21)1.1 (0.92,1.31)2.99 (1.7,5.26)* *P value <0.0006, accounting for multiple comparison. Conclusions: 1 year after SLE diagnosis, more than half of patient with LQ deteriorated. Patients in RR were more likely to deteriorate than improve. Plaquenil showed great promise in improving survival. Disclosure of Interest: None declared DOI: 10.1136/annrheumdis-2016-eular.4493Citation: Annals of the Rheumatic Diseases, volume 75, supplement 2, year 2016, page 562Session: SLE, Sjögren's and APS - clinical aspects (other than treatment) (Poster Presentations )