The inflammatory disease-associated variants in IL12B and IL23R are not associated with rheumatoid arthritis

There is increasing evidence that common genetic variants can contribute to general immune dysregulation and susceptibility to noninfectious inflammatory diseases. The cytotoxic T-lymphocyte-associated protein 4 (CTLA4) has been associated with Graves disease, type 1 diabetes (T1D), Addison’s disease, celiac disease and rheumatoid arthritis (RA) with odds ratios (ORs) ranging from 1.1 to 1.5 (1). In addition, a missense SNP, rs2476601, in the intracellular tyrosine phosphatase, PTPN22, has been found to be a risk factor (OR = 1.4–2.0) for a number of autoimmune diseases including RA, T1D, autoimmune thyroid disease and systemic lupus erythematosis (SLE) (1) while variants in the third intron of the transcription factor, STAT4, have been associated with risk to both RA and SLE (OR = 1.2–1.8) (2). Finally, the recent plethora of whole genome association (WGA) publications suggests common variants in other genes such as PTPN2 and IL2RA may contribute to risk of multiple noninfectious inflammatory diseases (3,4).

We recently found that common haplotypes in both IL12B and IL23R, marked by the IL12B SNPs rs3212227 and rs6887695 and the IL23R SNPs rs7530511 and rs11209026, were associated with psoriasis (odds ratios 1.4–1.5) (5) and association of these genes with psoriasis has now been replicated in another studies (6,7). Independently, Duerr and colleagues (8), and subsequently others (3,9), have shown that the IL23R SNP, rs11209026, which encodes an arginine to glutamine change at position 381, was also strongly associated with inflammatory bowel disease (IBD) (OR = 2.2–3.8). Interestingly, the psoriasis-associated IL12B SNP, rs6887695, has also been associated with Crohn’s Disease (CD) (9) and a large association scan of 14,500 nonsynonymous SNPs has identified the IL23R missense SNP, rs11209026, as a risk factor for ankylosing spondylitis (AS) (10).

IL12B and IL23R encode proteins that are members of the IL-12 superfamily of cytokines (reviewed in 11). IL12B encodes the IL-12p40 subunit of both the IL-12 and IL-23 cytokines and insertions and deletions in this gene have been associated with a rare recessive form of Mendelian susceptibility to mycobacterial infection while IL23R encodes one of the two subunits of the IL-23 receptor. The IL-12 and IL-23 cytokines, which play critical and unique roles in bridging the innate and adaptive immune systems, are produced primarily by activated dendritic cells and macrophages in response to microbial stimulation. IL-12, a pro-inflammatory cytokine, plays a central role in promoting the differentiation of naive CD4⁺ T cells into mature interferon-γ producing T-helper (T_H1) effector cells and it is a potent stimulus of natural killer (NK) and CD8⁺ T cells to produce IFN-γ. IL-23, on the other hand, is required for the generation of effector memory T cells and drives the expansion and maintenance of the newly defined ThIL-17 T cells that appear to play a fundamental role in autoimmunity.

The aim of this study was to determine whether the IL12B and IL23R SNPs associated with psoriasis and other noninfectious inflammatory diseases were also associated with RA and/or specific subphenotypes of this disease by genotyping the four implicated SNPs (rs3212227, rs6887695, rs7530511 and rs11209026) in three large, independent, white RA case-control sample sets (1732 cases/2502 controls). All cases included in this study were white, met the 1987 American College of Rheumatology diagnostic criteria (12) and have been described in detail elsewhere (13,14). Informed written consent was obtained from every subject and national and/or local institutional review boards approved all protocols and recruitment sites. Briefly, Sample Set 1, obtained by Genomics Collaborative, Inc. (GCI), consisted of 475 white rheumatoid factor (RF) positive RA patients and 475 individually matched controls (matched on gender, age +/− 5 years and grandparental country of origin) from North America. Sixty-six percent were female and the average age of onset was 47 years. Sample Set 2, obtained by the North American Rheumatoid Arthritis Consortium (NARAC), consisted of 661 unrelated RA patients (536 RF-positive and 125 RF-negative) from white multiplex families collected from across North America and 1322 healthy white control individuals from the New York Cancer Project (NYCP) matched two controls to one case for sex, age (decade of birth) and ethnicity (grandparental country/region of origin). The average age of disease onset was 38.6 years and 81% were female. Sample Set 3, from the Leiden University Medical Centre included 596 unrelated, white Dutch patients and 705 unrelated white controls. Complete phenotype data were not available on all the Dutch patients; however, 65% (362 of 558) were female and 72% were RF-positive (317 of 440). The average age of onset was 54.6 years (data was available for 306 patients). Information on carriage of the PTPN22 C1858T risk allele was available for all three sample sets. Complete information on HLA-DRB1 shared epitope (SE) status was available for sample sets 1 and 2.

Sample set 1 was genotyped using a multiplexed conventional polymerase chain reaction followed by a flow cytometry-based oligonucleotide ligation assay (15); sample sets 2 and 3 were genotyped for all four SNPs using kinetic PCR with allele-specific primers (16); (primer and probe sequences are available upon request). Genotypes were automatically called using custom software followed by hand curation without knowledge of case-control status. Previous cross-validation studies of these two platforms suggest a genotyping concordance of greater than 99.8% (5). Carriage of the HLA-DRB1 SE was determined as previously described (13).

An exact test of Hardy Weinberg Equilibrium (HWE) on the genotypic data, performing the analysis separately for cases and controls in each sample set, provided no evidence for deviation from HWE at any of the four SNPs tested (all P-values≥0.05) (Table 1). Case and control allele frequencies for all four SNPs were similar across all three studies and in agreement with control frequencies from three white, North American psoriasis case-control sample sets (5). No significant association between any of the four SNPs and RA was detected by either allelic or genotypic analyses (Table 1). Stratifying by the presence of RF, gender, age of onset, carriage of HLA-DRB1 SE positive alleles or carriage of the known PTPN22 1858T risk allele also revealed no consistent evidence for association across sample sets (data not shown).

Table 1.

Association of the IL12B and IL23R SNPs with RA

			Genotypes					Allelic (1 vs 2)		Genotypic
SNP (allele 1/allele 2)	Sample Set	case/ control	11	12	22	HWE P1	MAF	OR (95% CI)2	P3	P4
rs3212227 (C/A)	1	case	27	142	301	0.069	0.209
		control	23	145	306	0.316	0.201	1.04 (0.84–1.31)	0.732	0.830
	2	case	27	195	437	0.374	0.189
		control	61	433	823	0.679	0.211	0.87 (0.74–1.03)	0.111	0.248
	3	case	24	169	398	0.263	0.184
		control	30	191	478	0.055	0.180	1.03 (0.84–1.26)	0.798	0.871
	combined							0.96 (0.86–1.07)
rs6887695 (C/G)	1	case	52	194	223	0.338	0.318
		control	51	194	228	0.336	0.313	1.02 (0.84–1.24)	0.843	0.976
	2	case	58	278	323	0.926	0.299
		control	137	578	601	0.950	0.324	0.89 (0.77–1.03)	0.119	0.284
	3	case	58	237	296	0.299	0.299
		control	66	262	371	0.050	0.282	1.09 (0.91–1.29)	0.360	0.558
	combined							0.98 (0.89–1.08)
rs7530511 (T/C)	1	case	9	120	342	0.854	0.146
		control	6	101	368	1.000	0.119	1.27 (0.97–1.66)	0.078	0.210
	2	case	6	151	501	0.205	0.124
		control	14	267	1037	0.580	0.112	1.12 (0.91–1.38)	0.269	0.383
	3	case	11	144	436	0.823	0.140
		control	11	182	508	0.243	0.146	0.96 (0.77–1.20)	0.735	0.243
	combined							1.12 (0.94–1.33)
rs11209026 (A/G)	1	case	3	54	414	0.423	0.064
		control	0	60	415	0.243	0.063	1.01 (0.70–1.46)	1	0.128
	2	case	7	88	563	0.100	0.078
		control	4	182	1132	0.305	0.072	1.08 (0.84–1.39)	0.562	0.125
	3	case	1	84	505	0.194	0.073
		control	2	77	621	1	0.060	1.28 (0.94–1.75)	0.127	0.227
	combined							1.09 (0.96–1.25)

¹Calculated using Weir's exact test for Hardy Weinberg equilibrium (HWE).

²Combined ORs were calculated using a Mantel-Haenszel common OR. Results are reported for the minor allele.

³Calculated using Fisher's exact test.

⁴Calculated using William's-corrected G-test.

Power to detect a disease model with a continuity-corrected Mantel-Haenszel test (17) on allelic data was determined for each SNP using a Monte Carlo simulation (10,000 replicates for each data point) with the significance level set at alpha = 0.05 for all runs. Simulations were performed to estimate the disease model allelic OR necessary to achieve 80% power given the observed allele frequencies and sample sizes in our RA data sets. These results indicate we had 80% power to detect allelic odds ratios of 1.17 for rs3212227, 1.14 for rs6887695, 1.21 for rs7530511 and 1.27 for rs11209026.

Two SNP haplotypes were estimated for each gene (Table 2) using the Haplo.Stats package (18). Case and control haplotype frequencies for both genes were similar across all three samples sets and in agreement with control frequencies estimated from three white North American psoriasis case-control sample sets (5). No significant association was observed between these predicted IL12B and IL23R haplotypes and RA.

Table 2.

Two marker haplotypes for the IL12B and IL23R genes¹

a. IL12B
		Sample Set 1				Sample Set 2				Sample Set 3
Haplotype		Global P = 0.909				Global P = 0.327				Global P = 0.773
rs3212227	rs6887695	No. (Frequency) in				No. (Frequency) in				No. (Frequency) in
		Case (n=475)	Control (n=475)	OR	P	Case (n=661)	Control (n=1322)	OR	P	Case (n=596)	Control (n=705)	OR	P
A	G	600 (0.635)	604 (0.644)	0.98	0.731	874 (0.661)	1668 (0.632)	1.14	0.074	788 (0.668)	958 (0.683)	0.93	0.399
A	C	146 (0.155)	146 (0.155)	1.00	0.984	198 (0.150)	416 (0.158)	0.94	0.573	176 (0.149)	192 (0.137)	1.10	0.369
C	C	152 (0.162)	152 (0.161)	1.00	0.916	196 (0.149)	438 (0.166)	0.88	0.143	176 (0.149)	200 (0.143)	1.05	0.683
C	G	44 (0.048)	38 (0.040)	1.16	0.472	54 (0.040)	118 (0.045)	0.91	0.527	40 (0.034)	52 (0.037)	0.91	0.741

b. IL23R
		Sample Set 1				Sample Set 2				Sample Set 3
Haplotype		Global P = 0.162				Global P = 0.559				Global P = 0.286
rs7530511	rs11209026	No. (Frequency) in				No. (Frequency) in				No. (Frequency) in
		Case (n=475)	Control (n=475)	OR	P	Case (n=661)	Control (n=1322)	OR	P	Case (n=596)	Control (n=705)	OR	P
C	G	746 (0.789)	772 (0.822)	0.81	0.108	1056 (0.798)	2162 (0.818)	0.88	0.182	930 (0.787)	1116 (0.796)	0.95	0.528
T	G	138 (0.147)	106 (0.114)	1.34	0.057	164 (0.124)	290 (0.110)	1.15	0.252	166 (0.140)	204 (0.146)	0.96	0.712
C	A	60 (0.064)	56 (0.061)	1.07	0.990	102 (0.078)	184 (0.070)	1.12	0.527	86 (0.073)	82 (0.058)	1.26	0.117
T	A	0 (0)	4 (0.003)	NC	NC	0 (0)	6 (0.002)	NC	NC	0 (0)	0 (0)	NC	NC

¹The Haplo.Stats package (18) was used to estimate haplotypes and test for association with disease status. P-values and ORs are for the indicated haplotype compared to all others.

Lack of association of the IL12B 3’UTR SNP, rs3212227, with RA has been reported in case-control studies of Greeks (179 cases / 159 controls), British (258 cases / 255 controls) and Spanish (545 cases / 393 controls) (19,20). Our results confirm the lack of association of rs3212227 with RA and extend these findings by showing that the psoriasis-associated 5’ IL12B SNP, rs6887695, and the haplotypes predicted by it and rs3212227 are also not associated with RA. In addition, Orozco and colleagues (21) recently characterized eight IL23R SNPs, including rs11209026, in 322 RA cases and 243 healthy controls from Southern Spain and did not observe any of these SNPs to be associated with RA predisposition. Haplotypes were not tested, nor was rs7530511 and, according to the CEU HapMap data, none of the eight genotyped SNPs were in substantial LD with this SNP (all r² values ≤0.153). Consequently, our data confirm lack of association of the IL23R R381Q missense SNP with RA and extends these results by showing that rs7530511 and the rs7530511-rs11209026 haplotypes do not appear to be associated with RA.

Since inflammatory diseases associated with the same genetic variant may share a common underlying mechanism, a thorough understanding of the diseases associated with a particular variant may shed light on disease pathogenesis. Consequently, the finding that a subset of psoriasis-associated IL12B and IL23R SNPs are also associated with CD and AS led us to carefully evaluate the role of these same SNPs in other inflammatory diseases. We recently reported the results of a well-powered, family-based association analysis of 910 multiple sclerosis (MS) nuclear families (3132 individuals) which provided no evidence that any of these four SNPs were associated with MS risk, even after the data were stratified by sex of the patient, presence/absence of the risk HLA-DRB1*1501 allele, and disease severity (22). Our analyses of three RA case-control sample sets (1732 cases / 2502 controls) presented here suggest that there is also no evidence of association between these same SNPs/haplotypes and RA, even after the data were stratified for sex of the patient, age of disease onset, presence/absence of the risk HLA-DRB1 SE, presence/absence of the risk PTPN22 1858T allele, and RF-status. These studies were well powered to detect modest effect sizes allowing us to rule out strong RA effects at these four SNPs; however, it is still possible that these SNPs could play a minor role in risk for RA. This will be uncovered by using much larger sample sets.

These disease-association patterns are interesting in light of the recent proposal by McGonagle and McDermott (23) that all noninfectious inflammatory diseases lie on a spectrum from autoimmune, those mediated by the adaptive immune system, to autoinflammatory, those mediated by the innate immune system, and that each disease can be defined by the relative contributions of each. The inflammatory diseases associated with the IL12B and IL23R variants reported here are all lie toward the autoinflammatory end of the spectrum; CD and AS are thought to be classic autoinflammatory diseases while psoriasis is proposed to have a mixed pattern of both autoimmune and autoinflammatory characteristics. RA, on the otherhand, is primarily autoimmune in nature. Interestingly, RA and a number of other classic autoimmune diseases, such as SLE and T1D, are associated with the R620W variant in PTPN22 while CD, AS, psoriasis and other autoinflammatory diseases are not (1).

In summary, the IL12B and IL23R SNPs and haplotypes associated with risk for psoriasis and other inflammatory diseases do not appear to play a major role in RA risk in our data sets suggesting that these variants in the IL-12/IL-23 pathway may not differentially influence RA pathogenesis. Analyses of the role of these genetic variants in additional inflammatory disorders should provide hints about the underlying mechanisms of these complex diseases as well as a framework to begin to understand why certain inflammatory diseases often show familial aggregation.