Discussion
Our study demonstrates associations between high genetic burden in B cell-related pathways, autoantibody production and LN in SLE, providing clues on a potential tool to assess future risk of organ damage of patients with SLE.
Important findings include first, increased prevalence of SLE in individuals with high B cell-related genetic risk. It has previously been shown that the weighted effect of multiple SLE-related SNPs likely plays a more important role in SLE pathogenesis than any individual high-risk genes alone.8 It is therefore not surprising to find a smaller subset of SLE risk loci to have a similar effect on disease risk. However, the limitation to genes in B cell-related pathways adds important information on what pathways might be dominating in pathogenesis for some patients.
Second, it was shown that patients with SLE with a high SLE B cell PRS were more likely to fulfil the immunological disorder criterion and to present with anti-dsDNA antibodies. This supports the hypothesis that generation of autoreactive antibodies is promoted by genetic aberrations in B cells to a significant extent. A variety of B cell abnormalities have already been described in SLE. Notably, it has been demonstrated that patients with SLE display imbalance in B cell populations in peripheral blood, altered function in immunosuppressive regulatory B cells and expansion in subsets of autoreactive effector B cells.11 12 Our findings are consistent with previous knowledge, and suggest a method to identify patients with the highest degree of B cell-related genetic burden, and therefore the highest risk of developing autoantibody-related organ damage.
No significant association was observed between the B cell PRS and other autoantibodies, but a trend, with ORs above 1, was shown for anti-Sjögren’s syndrome-related antigen A (SSA), anti-Sjögren’s syndrome-related antigen B (SSB) and anti-β2-glycoprotein I antibodies. This result could be related to low power in the analysis, but could also be due to a specific effect of the PRS on anti-dsDNA antibody production. Compared with SSA and SSB antibodies, titres of anti-dsDNA antibodies vary over time. One could therefore speculate that gene variants included in the PRS affect B cells and plasma cells, increasing production of anti-dsDNA antibodies above the level of detection. However, further studies are needed to clarify if this is the case.
When comparing the effect of the B cell PRS with the overall non-HLA genetic risk for SLE on anti-dsDNA antibody development, we found the OR to be nominally higher for the SLE PRS. However, when removing the B cell SNPs from this PRS, the nominal OR was lower than that of the B cell PRS. This in spite of the B cell PRS including a considerably smaller number of SNPs (20 vs 95). These results suggest that the average B cell-associated SLE locus may contribute more to development of anti-dsDNA antibodies than the average SLE risk SNP in general. But also, as can be expected, that other SLE risk loci are important for the production of autoantibodies.
Third, HLA risk variants HLA-DRB1*03:01 and HLA-DRB1*15:01 were shown to augment the effect of a high SLE B cell PRS in anti-dsDNA antibody development. The underlying mechanism of HLA haplotype in context of SLE susceptibility is yet unknown. However, our results indicate that high genetic burden related to B cells plays a more central role for patients with certain HLA types, perhaps because the combination of increased B cell reactivity and HLA-mediated antigen presentation predisposes for dsDNA antibody production, more than either risk factor alone.
Here, we also demonstrated that patients with HLA-DRB1*03/15 +/+ and high SLE B cell PRS had higher prevalence of low complement levels. Reduction in C3, C4 and CH50 levels is well described in SLE, and has earlier been inversely associated with rising levels of dsDNA antibodies.39 40 Our results could be a reflection of a subgroup of patients’ genetic predisposition to produce anti-dsDNA antibodies, subsequent formation of immune complexes and complement consumption.
Although a significant association between high SLE B cell PRS and LN could not be shown here, we demonstrate a significantly higher prevalence of LN in patients with high PRSs related to B cell activation. One explanation to this might be the relatively larger contribution of the BANK1 SNP rs10028805 in the SLE B cell activation PRS, compared with the SLE B cell PRS. We have previously shown variants in BANK1 to be associated with LN in a cohort partially overlapping with the present study population.20 The B cell activation PRS was not as strongly associated with nephritis as presence of dsDNA antibodies. This might not be surprising given the well-known strong association between these antibodies and LN.15 Further studies are needed to clarify the role of the B cell activation PRS, available early in disease, in nephritis prediction.
Strengths of this study include a large cohort and extensive clinical information with well-defined SLE classification criteria. Another advantage is the long-time follow-up, which however has allowed for the use of different methods and cut-off levels for measurement of autoantibodies. Limiting analyses to women allowed for analysis of a cohort with as much genetic similarities as possible. However, the limitation to female gender and European ancestry was also a weakness, as the results are only applicable to this group. The clinical phenotype of SLE varies in populations of different descent, trending towards a more active disease with more organ damage and higher mortality in patients of African ancestry, compared with patients of white populations.41 The same is known for males with SLE, who are known to more often develop LN and to have a generally more aggressive disease.42
With increasing knowledge of genetic and immunological pathways influencing LN pathogenesis, it is likely that the future of LN management lies in precision medicine. An important step forward is the identification of biomarkers or instruments enabling early recognition of patients genetically predisposed to develop LN, which could motivate closer monitoring or even pre-emptive therapy.43 Assessing B cell PRSs could be one important component in developing such instruments. Future areas of research would be to verify the results presented here in an independent cohort. Further, it will be important to examine if and how the B cell PRSs influence gene expression in cell subsets such as B cells and cells in kidney tissue. It would also be interesting to investigate if it could be used to predict clinical response to B cell targeting therapy in SLE. To conclude, assessing B cell polygenic risk may be important in order to determine immunological pathways influencing SLE. However, further studies are needed to clarify the relationship between genetics and clinical phenotype in SLE, and to improve PRS performance.