Background Systemic lupus erythematosus (SLE) is a clinically heterogeneous disorder with a waxing and waning clinical course. Reliable markers of clinical disease flare have been difficult to identify, impeding the selection of optimal therapies, which leads to suboptimal disease control and cumulative organ damage.
Objectives Our goal was to identify immune markers associated with impending vs delayed flare in SLE patients in the absence of confounding immunosuppressive medications.
Methods As part of the Biomarkers of Lupus Disease (BOLD) study, 41 SLE patients with non-organ-threatening, moderately severe disease activity were enrolled, background immunosuppressants (IS) discontinued, intramuscular steroids given until disease suppression, and followed until clinical flare. Patients met ≥4 ACR SLE criteria and had SLEDAI ≥6 or BILAG ≥2 B or 1 A scores at baseline. Samples were procured at baseline, time of disease suppression, and serially until flare (defined as requiring new treatment and ≥1 new BILAG B or ≥4 point SLEDAI increase). Cellular immunophenotyping markers and 52 soluble mediators were measured using xMAP multiplex technology or sandwich ELISA. Random forest was used to generate predictive models and applied as described .
Results Forty SLE patients flared within 24 weeks, with 21 flaring within 60 days (early) and 13 flaring later than 90 days (late) after stopping background IS. Patients who flared early were more likely to be of African-American descent, while Native American patients were more likely to flare late. At baseline those who flared early had higher levels of bicarbonate, increased frequencies of monocytes and neutrophils expressing the activated conformation of CD11b, monocytes (p=0.028) and neutrophil (p=0.002), as well as greater frequency of activated naïve B cells. SLE patients who flared late had higher baseline plasma levels of IL-1RA (p=0.03) and TNFRI (p=0.04). At the flare visit this group had higher levels of BLyS (p=0.01), IL-7 (p=0.03), and IFNg (p=0.04). Predictive models were generated and tested using random forest models. The best prediction model for the course of flare based on three immunophenotyping parameters achieved 87% accuracy with a positive predictive value of 0.78 (95% CI, 0.52 – 0.94) and a negative predictive value of 0.80 (0.44 – 0.97). The model indicates that absence of the activated monocyte population (activated CD11bhi) and the activated naïve B cell subpopulation (CD86hi) led to a drastically reduced likelihood of an early flare.
Conclusions SLE patients who flare earlier after withdrawal of ineffective immuno-suppressants and transient steroid treatment have increased measures of innate and/or adaptive immune system activation, while SLE patients who flare later have increased evidence of regulatory pathway engagement through shedding of IL-1RA.
Genuer R, et al. Pattern Recogn Lett. 2010;31(14):2225-36.
Disclosure of Interest : J. Guthridge: None declared, R. Lou: None declared, S. Kamp: None declared, M. Munroe: None declared, K. Bean: None declared, S. Macwana: None declared, S. Sridharan Employee of: Pfizer Inc, J. Merrill Grant/research support: Pfizer Inc, Consultant for: Pfizer Inc, J. James: None declared