Background It is important to identify biomarkers of SLE disease activity for patient care, research, and drug development. Many gene transcripts and proteins in blood or urine have been observed to correlate with disease activity in SLE. However some observed associations might be spurious, due to confounding by correlation with other biomarkers or patient characteristics
Objectives In this study, we explored the relationship between six proposed biomarkers and SLE activity over a 1 year period while controlling for potential confounding variables.
Methods At an initial visit, two proteins and four gene transcripts or signatures were measured in 280 SLE patients. Levels of the BAFF gene transcript, plasma cell (PC) gene signature, IFN gene signature, and an LDG-associated Neutrophil gene signature were assessed in PAXgene-preserved peripheral blood by global microarray and qPCR. For proteins, BAFF was measured in serum and TWEAK in urine (both by ELISA). Disease activity during the next year was quantified by SELENA-SLEDAI modified to exclude complement and dsDNA. Repeated measures linear regression models were fit to determine which markers were predictive of disease activity over a one year period, controlling for age, race, sex, and other markers. Non-linear relationships between biomarker levels and disease activity were also explored.
Results In univariate analyses, all markers analyzed except BAFF protein were significantly associated with future disease activity. After controlling for race, the PC signature was no longer significantly associated. After controlling for BAFF mRNA levels, the IFN signature was no longer significantly associated. Controlling for sex, race, and other biomarkers we found that: 1) a 1 standard deviation (SD) increase in BAFF was associated with a mean SLEDAI increase of 0.26 in the follow-up (p=0.0034), 2) those patients within the top 15% of the Neutrophil gene signature expression had a 0.66 higher mean SLEDAI during follow-up (p=0.0056), and 3) the relationship between the SLEDAI score and urinary TWEAK protein was constant until the 85th percentile of TWEAK after which a 1 SD increase in TWEAK was associated with a 0.58 increase in mean SLEDAI (p=0.0006). In a similar analysis focusing on renal disease activity, a 1 SD increase in the IFN signature was associated with a mean renal SLEDAI increase of 0.11 (p=0.04). The Neutrophil signature remained significant at a similar level as for overall disease activity, and the relationship between renal SLEDAI score and urinary TWEAK was linear with a 1 SD increase in TWEAK being associated with a 0.25 increase in mean renal SLEDAI (p<0.0001).
Conclusions BAFF gene transcript, LDG-associated neutrophil gene signature, and high levels of urinary TWEAK appear to be independently and additively associated with disease activity. Our results suggest that the association between IFN and overall disease activity is due to the association between IFN and BAFF. Our results also suggest that an observed association between PC and disease activity is due to confounding by race. Thus, given that biomarkers are correlated with each other and other risk factors for disease, it is important to adjust for confounding when assessing biomarker/disease relationships.
Disclosure of Interest None declared