In this study we have utilised a large cohort of JIA cases and identified associations between UK JIA and polymorphisms in three genes: MVK, TNFRSF1A, and IL1A. These findings follow on from work performed by our group, showing associations between psoriatic JIA and other HPF syndrome genes, and work previously showing associations between SoJIA and the IL1 ligand and IL1 receptor gene clusters [7, 18].
After applying a Bonferroni correction for the number of loci studied (p<0.0013), none of the loci associated with UK JIA would remain significant. However, this is a highly conservative correction method which assumes all tests are independent, as it does not take into account correlations between SNPs. We have thus taken the approach of trying to validate significant findings in an independent cohort.
For two of the four significantly associated SNPs in the UK cohort we had validation data from a US cohort. One SNP showed no association in this cohort and the other SNP showed association but in the opposite direction. The US cohort comprises JIA cases from only three ILAR subtypes: RF negative polyarthritis, persistent and extended oligoarthritis subtypes, whereas the UK cohort encompasses all seven of the ILAR JIA subtypes. Analysis of the UK dataset stratified by ILAR subtype for the rs7957619 and rs11836136 SNPs suggests the association with JIA overall is driven largely by the systemic, together with the RF positive polyarthritis and psoriatic arthritis subtypes (Figure 2a and b). Analysis with just the RF negative polyarthritis and oligoarthritis subtypes showed no association for either rs11836136 (ptrend=0.3) or rs7957619 (ptrend=0.12). Conversely, analysis of the UK dataset with all the other subtypes (systemics, RF positive polyarthritis, enthesitis related arthritis and psoriatic JIA) showed association with rs11836136 (ptrend=0.005) and rs7957619 (ptrend=0.02). Therefore, this represents evidence for subtype specific associations. This is an interesting finding in itself as it provides further genetic evidence that the SoJIA cases are different to the other subtypes. Further investigations into the genetics of individual subtypes of JIA are required however, power is a big issue as stratification by subtype inevitably leads to small sample sizes. The lack of, or weak, association of these SNPs in the oligoarthritis and RF negative polyarthritis subtypes in the initial cohort may have been due to a lack of power and so investigating another cohort of US JIA cases and controls was appropriate. This approach has strengthened the evidence that these finding represent subtype specific effects. Independent validation of the finding in SoJIA is still required in other datasets.
The MVK gene encodes for mevalonate kinase, an enzyme that plays a key role in steroid synthesis. It catalyses the conversion of mevalonic acid to 5-phosphomevalonic acid, and mutations in this gene are responsible for both the hyper-IgD syndrome (HIDS), and the more severe mevalonic aciduria [14, 23]. The pathogenesis of HIDS is not completely clear, but despite very high levels of IgD seen in this disease, it is thought this is unlikely to exert a direct pathological effect. More likely it is a reduction in levels of isoprenoids, anti-inflammatory compounds produced downstream from MVK in the steroid synthesis pathway that causes the proinflammatory phenotype seen in HIDS . Experimental models have shown that lack of isoprenoids in HIDS causes an increase in the secretion of the proinflammatory cytokine IL-1β [24, 25]. The rs7957619 SNP associated with UK JIA is exonic and changes an amino acid (from Serine to Asparagine). Once replication of the genetic association has been achieved functional analysis of this coding SNP should be undertaken to determine if this variation alters the quality or quantity of the resulting MVK protein and if any downstream effects on steroid synthesis results.
We also found association of a SNP close to TNFRSF1A. TNFSR1A encodes the p55 (type-1) TNF receptor (TNFR1), which plays a vital role in regulating the innate immune system via TNF-α signaling, resulting in downstream cytokine secretion, NF-kB activation, and apoptosis. The rs2228576 SNP lies 3’ of TNFRSF1A, within exon 13 of SCNN1A. The SNP results in a Threonine to Alanine amino acid change. The SCNN1A gene itself, which is primarily involved in sodium and water transport across epithelia, seems an unlikely gene to have functional relevance in JIA. However, once replicated, the ability of rs2228576 to regulate TNFSR1A function could be determined using chromosome conformation capture techniques [26, 27]. This would have direct relevance to JIA disease pathogenesis. We found association of a SNP in the IL1A gene with the total UK JIA dataset. Stratification of this SNP by ILAR subtype however, again showed that this was largely driven by association in the SoJIA subtype. It is important to point out that there is almost complete overlap between the SoJIA cases studied here and by Stock et al.. Therefore, the IL1 ligand and receptor SNP associations described with SoJIA herein are not independent validation of Stock’s findings. However, our data does provide evidence to suggest that the IL1 ligand and receptor SNP associations are subtype specific, being associated primarily with SoJIA, and possibly also with the psoriatic JIA subtype. Interestingly, there was a trend also towards significance with the MVK rs7957619 (p=0.08) and rs11836136 (p=0.06) SNPs with psoriatic JIA. The numbers in this subtype are small and independent replication is required however, it is intriguing that trends towards association with additional autoinflammatory disease loci and psoriatic JIA subtype are occurring, especially as we have previously reported significant association with SNPs in the autoinflammatory disease genes NLRP3, NOD2, MEFV and PSTPIP1 with psoriatic JIA .
Magitta et al. previously found no association with two SNPs in NLRP1 when 505 Norwegian JIA cases were studied . We have conducted a comprehensive SNP analysis of the whole NLRP1 gene, and its 5’ and 3’ flanking regions and found no association with UK JIA.
Although JIA is the most common rheumatic disease of childhood, it is nevertheless extremely rare, making large cohorts difficult to collect. This makes small sample sizes a continuing problem across all JIA research, especially research focused on the less common subtype presentations. Here we have used the largest cohort of JIA available in Europe, and utilised data from the Welcome Trust Case Control Consortium to provide control genotype data, to maximise study power. We have identified association to SNPs in three genes that appear to be specific with SoJIA, highlighting that there are genetic as well as clinical differences between SoJIA and the other JIA subtypes.