Association of a rare variant of the TNFSF13B gene with susceptibility to Rheumatoid Arthritis and Systemic Lupus Erythematosus

A rare variant (BAFF-var) of the tumor necrosis factor superfamily 13b (TNFSF13B) gene has been recently associated with multiple sclerosis (MS) and systemic lupus erythematosus (SLE). The aim of this study was to investigate the association between TNFSF13B BAFF-var and susceptibility to rheumatoid arthritis (RA) and replicate that association in SLE. 6,218 RA patients, 2,575 SLE patients and 4,403 healthy controls from three different countries were included in the study. TNFSF13B BAFF-var was genotyped using TaqMan allelic discrimination assay. PLINK software was used for statistical analyses. TNFSF13B BAFF-var was significantly associated with RA (p = 0.015, OR = 1.21, 95% CI = 1.03–1.41) in the Spanish cohort. A trend of association was observed in the Dutch (p = 0.115) and German (p = 0.228) RA cohorts. A meta-analysis of the three RA cohorts included in this study revealed a statistically significant association (p = 0.002, OR = 1.24, 95% CI = 1.10–1.38). In addition, TNFSF13B BAFF-var was significantly associated with SLE in the Spanish (p = 0.001, OR = 1.41, 95% CI = 1.14–1.74) and the German cohorts (p = 0.030, OR = 1.86, 95% CI = 1.05–3.28), with a statistically significant p-value obtained in the meta-analysis (p = 0.0002, OR = 1.46, 95% CI = 1.09–2.32). The results obtained confirm the known association of TNFSF13B BAFF-var with SLE and, for the first time, demonstrate that this variant contributes to susceptibility to RA.

Rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE) are complex autoimmune diseases influenced by both genetic and environmental factors 1,2 . Genome-wide association studies (GWAS) have revealed more than 100 independent signals for RA and 50 for SLE associated with these diseases 3,4 . In a recent study Steri et al. 5 identified an association signal in the TNFSF13B gene which was associated with multiple sclerosis (MS) and SLE. TNFSF13B encodes tumor necrosis factor superfamily member 13b (also known as B-cell activating factor (BAFF)). BAFF is a cytokine that is primarily produced by monocytes and neutrophils and plays a crucial role in B-cell homeostasis and the regulation of B-cell maturation, differentiation and survival 6 . This cytokine plays a particularly important role in the pathogenesis of SLE, to such an extent that the first targeted therapy approved for SLE was Belimumab, a monoclonal antibody targeting human BAFF [7][8][9] . The causal variant identified by Steri et al. 5 for this association, "BAFF-var", is an insertion-deletion in which five nucleotides are replaced by one (GCTGT > A), being this A the risk allele. Persons carrying the A allele have an increased risk for the disease. This risk allele results in a shorter transcript that escapes microRNA inhibition thus leading to an increase in the production of soluble BAFF. Besides, it has been observed that TNFSF13B BAFF-var is strongly associated with augmented levels of total IgG and IgM and with reduced monocyte counts 5 . Thus, TNFSF13B BAFF-var is considered a shared genetic risk variant for autoimmune diseases based on its association with MS and SLE, and it plays an important role in autoimmune processes 5 . Considering that replication of previous results is of vital importance for the correct development of scientific knowledge, we performed a replication study to corroborate the association of TNFSF13B BAFF-var with susceptibility to SLE. Otherwise, BAFF system dysregulation is involved in the pathogenesis of RA 1,10 and abnormal levels of BAFF have been detected in serum, synovial fluid and saliva from RA patients 11 . It is noteworthy that more than 10 shared risk loci have been identified for SLE and RA 12,13 . Taking into considerations above, we aimed to assess for the first time the potential association of TNFSF13B BAFF-var with RA in patients from Spain, the Netherlands and Germany. Genotyping. Genomic DNA was isolated from peripheral blood samples using standard salting-out techniques. TNFSF13B BAFF-var was genotyped using a TaqMan allelic discrimination custom assay (assay ID: AH0JGPG) (Applied Biosystems, Foster City, California, USA) on an ABI 7900HT Fast Real-Time PCR System (Thermo Fisher). Genotyping call rate was >95% in all sample sets. Statistical analysis. All the statistical analyses were carried out with PLINK v1.9 (http://pngu.mgh.harvard. edu/purcell/plink/) 17 . For all groups of individuals, Hardy-Weinberg equilibrium was determined at a significance level of 0.01. Differences between allele and genotype frequencies in individuals were determined by χ 2 test. Odds ratios (ORs) and 95% confidence intervals were calculated according to Woolf 's method. Meta-analyses were performed with inverse-variance method under a fixed-effects model. Heterogeneity of the ORs across cohorts was assessed using both Cochran's Q test and inconsistency index (I 2 ). The statistical power of our study was calculated by using Power Calculator for Genetic Studies 2006 (CaTS) software (http://www.sph.umich.edu/csg/ abecasis/CaTS/) 18 , under an additive model. The statistical power of the study for determining significance at a p value of 0.05 is shown in Supplementary Table S1.

Results
First, in order to evaluate the association between the TNFSF13B gene variant (BAFF-var) and susceptibility to RA, a comparative analysis of allelic and genotype frequencies in RA patients vs. healthy individuals was performed in the Spanish, Dutch and German populations. Results are shown in Table 1. The minor allele frequency (A) of TNFSF13B BAFF-var was higher in RA patients in the three populations included in this study. This difference was statistically significant in the Spanish subgroup (p = 0.015, OR = 1.21, 95% CI = 1.03-1.41), and suggestive associations were observed in the Dutch (p = 0.115, OR = 1.47, 95% CI = 0.91-2.36) and German (p = 0.228, OR = 1.39, 95% CI = 0.81-2.36) subgroups. In addition, a statistically significant p value was obtained in the meta-analysis of the three RA populations included in this study (p = 0.002, OR = 1.24, 95% CI = 1.10-1.38) with low heterogeneity levels (heterogeneity q value = 0.64, I 2 = 0) ( Table 1).
Secondly, to try to replicate the results previously published by Steri et al. 5 , a comparative analysis of allelic and genotype frequencies in SLE patients vs. healthy individuals from the Spanish and German populations was also performed ( Table 2). The minor allele frequency (A) in TNFSF13B BAFF-var was statistically significantly higher in SLE patients than in controls in both, the Spanish (p = 0.0001, OR = 1.41, 95% CI = 1.14-1.74) and German (p = 0.030, OR = 1.86, 95% CI = 1.05-3.28) subgroups. Meta-analysis of the two SLE populations included in this study revealed a statistically significant p value (p = 0.0002, OR = 1.46, 95% CI = 1.09-2.32) with low heterogeneity levels (heterogeneity q value = 0.37, I 2 = 0) ( Table 2).

Discussion
We confirm the recently reported association of the TNFSF13B BAFF-var with SLE in two independent populations of Spanish and German origin. In addition, this is the first time that evidence is provided of an association between TNFSF13B BAFF-var and RA. These findings, together with the data reported by Steri et al. 5 , suggest that TNFSF13B BAFF-var is a common genetic risk factor for autoimmunity. SLE and RA mice models showed increased serum BAFF levels, and BAFF blockade reduced diseases manifestations 19,20 . Nevertheless, although clinical trials are being performed on BAFF in RA and SLE patients, only moderate results are emerging. An example of these modest results is the BAFF-inhibitor drug belimumab, which has been found to be only moderately effective in a small number of patients with RA and SLE 8,21 . Because RA and SLE are heterogeneous diseases with different biological subsets, the efficacy of different drugs could be divergent in specific patient subgroups. Therefore, our new findings could help in the advance of RA therapies, as patients stratified by TNFSF13B BAFF-var status may show a differential benefit from anti-BAFF therapies. In this sense, a recent study showed that a TNFSF13B genetic variant influenced response to B-cell targeted therapy (rituximab) in patients with RA 22 . Furthermore, B-cell-depleting therapies in patients carrying TNFSF13B BAFF-var would have a weaker effect due to a rapid resurgence of memory B cells induced by high soluble BAFF levels, which could increase the risk of inadequate response and/or relapse 23 .
Genome-wide association studies (GWAS) have identified multiple risk genetic factors associated with human autoimmune disease. Of note, GWAS are designed to detect only common genetic variants (>5%); however, low-frequency and rare variants may represent an important component of autoimmune risk and provide a key  insight into both novel and previously implicated immunological pathways that are disrupted in autoimmune diseases 24 . Thus, large genetic studies have identified low frequency variants with relevant functional significance associated with autoimmune diseases (e.g. TREX1 in SLE 25 and TYK2 in several autoimmune diseases 26,27 ). Our findings support the important role of rare variants, such as TNFSF13B BAFF-var, in understanding the unknown mechanisms of autoimmunity. For the first time, our results showed that allele A of TNFSF13B BAFF-var was associated with the risk of RA in the Spanish population. Trends of association were observed in the Dutch and German RA cohorts. This could be due to a low statistical power as the Dutch and the German cohorts sample sizes are relatively small as compared to the Spanish cohort, as well as the minor allele frequency is less common in northern Europe (2.02% in Germany and 1.91% in the Netherlands) than in southern Europe (4.2% in Spain). This north-south gradient in the minor allele frequency of TNFSF13B BAFF-var has been previously reported, with higher frequencies observed in southern Europe (5.7% in Italy and 4.9% in Spain) than in northern Europe (1.8% in United Kingdom and Sweden) 5 . However, the three RA cohorts showed higher levels of allele A in cases than in controls and statistically significant association was observed in the meta-analysis. Thus, further studies involving larger cohorts would be necessary to elucidate if this variant is associated with autoimmune risk in northern populations.
Regarding the SLE replication study, our results confirm the association of allele A of TNFSF13B BAFF-var with the risk of SLE in the Spanish and German populations. The association of the variant was found to be stronger in SLE than in RA and, in both cases, the allele A acts as a risk factor.
GWAS have revealed a significant shared genetic component in autoimmunity 28,29 . The association of TNFSF13B BAFF-var with RA and SLE observed in our study extends the number of overlapping genetic risk factors across different autoimmune diseases. In this regard, performing further studies to elucidate the specific association of TNFSF13B BAFF-var with other autoimmune diseases would be of great interest as new therapy approaches involving BAFF could emerge.
In conclusion, our results suggest that TNFSF13B BAFF-var plays an important role in susceptibility to RA, and confirm the association of this variant with SLE.