Statistics from Altmetric.com
Psoriatic arthritis (PsA) is a chronic inflammatory arthritis associated with the presence of psoriasis. Although the majority of PsA genetic risk loci identified also confer risk for psoriasis, the difference in heritability between the two diseases suggests that there remain uncovered risk loci that are associated with PsA but not psoriasis (herein called PsA-specific loci).1 Here we present an independent replication of a PsA-specific association to rs12044149 at the well-established psoriasis risk locus, IL23R, and confirm its independence of the previously reported psoriasis single-nucleotide polymorphism (SNP); rs9988642.2 ,3
Following quality control, genotype data were available for Spanish, Cretan, German and UK PsA cases (914) and controls (6945), independent of those tested previously2 for rs12044149 and rs9988642. All PsA case subjects were recruited from rheumatology clinics and were diagnosed by a rheumatologist. Genotype data were generated using the Life Technologies TaqMan chemistry on the QuantStudio platform. Data were also available for samples that had been previously genotyped. Case–control association testing was performed separately for each dataset using logistic regression in PLINK. This was followed by a meta-analysis of the summary statistics, using an inverse-variance fixed-effects model (table 1). The association of rs12044149 with PsA was replicated (p=4.03×10−6) and remained significant after including rs9988642 as a covariate (pcond=4.86×10−6). Meta-analysis of this data with that of our previous Immunochip study reached genome-wide significance (pmeta=4.76×10−20) in 2876 cases and 15 868 controls (table 1). For the previously reported psoriasis variant, rs9988642,3 only modest association was found with PsA in the replication meta-analysis (p=0.04), and did not reach genome-wide significance upon meta-analysis of the combined PsA dataset (p=4.61×10−4) (table 1).
We investigated differences between PsA and psoriasis by combining our data with a subset of the psoriasis WTCCC2 study,4 excluding patients with known PsA (1784 psoriasis cases, 5175 controls). Here, rs9988642 was significantly associated with psoriasis (p=1.0×10−7) and remained so after conditioning for rs12044149 (pcond=1.63×10−5). The effect estimates of rs12044149 for PsA and psoriasis were significantly different (p=2.0×10−3), when tested using multinomial logistic regression in Stata. Direct comparison of the PsA and psoriasis genotypes revealed a significant association with an increased risk of PsA with rs12044149 (p=4.52×10−4, OR=1.3). While we cannot exclude the possibility of undiagnosed PsA cases in the psoriasis group, their inclusion would have biased the results to the null hypothesis of no difference in the association statistics between PsA and psoriasis. These results support previously reported evidence that the association signals for the IL23R variants rs12044149 and rs9988642 are independent of each other and that the association to rs12044149 is specific to PsA.2 ,5
A Bayesian refinement method was applied to define credible SNP sets based on each effect, as described previously.2 These were localised to regulatory features using data from the Roadmap Epigenomics Project6 and the online genetic and epigenetic finemapping data portal7 (figure 1). Credible SNPs (n=13) for the PsA-specific associated SNP mapped to promoter and enhancer regions within memory CD8+ T cells, which we have previously reported to be critical for PsA.2 By contrast, credible SNPs based on the psoriasis association (n=7) did not overlap with regulatory elements and one, rs11209026, was found to cause a missense mutation, resulting in an Arg381Gln substitution.8 This could suggest the involvement of different functional mechanisms for the PsA and psoriasis associations.
In conclusion, we have replicated a PsA-specific association at the IL23R locus within an independent population, and confirm that the association with rs12044149 is distinct from the psoriasis IL23R variant, rs9988642. Currently, five robust PsA-specific loci have been identified, including HLA-B,2 chromosome 5q31,2 PTPN229 and TNFAIP3.5 Such associations can provide potential clinical benefits in disease-risk estimation and linking genetics with therapeutic targets.
We thank Arthritis Research UK for their support (grant ref 20385).
Funding This report includes independent research funded by the NIHR Manchester Musculoskeletal Biomedical Research Unit. AB-A is funded by a studentship from the MRC. HB and FB were supported by the German Federal Ministry of Education and Research (ArthroMark 01EC1401C) as were UH and AR (METARTHROS 01EC1407A).
Disclaimer The views expressed in the publication are those of the authors and not necessarily those of the NHS, the National Institute of Health Research or the Department of Health.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.