Polymorphisms in STAT‐4, IL‐10, PSORS1C1, PTPN2 and MIR146A genes are associated differently with prognostic factors in Italian patients affected by rheumatoid arthritis

Summary

Rheumatoid arthritis (RA) is a systemic autoimmune disease resulting in chronic inflammation of the synovium and consequent cartilage and bone erosion. RA is associated strongly with the presence of rheumatoid factor (RF), and consists of clinical subsets of anti‐citrullinated protein antibody (ACPA)‐positive and ‐negative patients. This study was designed to evaluate whether relevant single nucleotide polymorphisms (SNPs) associated with RA and other autoimmune disorders are related to RF, ACPA and clinical phenotype in a cohort of biologic drugs naive Italian RA patients; 192 RA patients and 278 age‐matched healthy controls were included. Clinical and laboratory data were registered. We analysed a total of 12 single nucleotide polymorphisms (SNPs) in signal transducer and activator of transcription‐4 (STAT‐4), interleukin (IL)‐10, psoriasis susceptibility 1 candidate 1 (PSORS1C1), protein tyrosine phosphatase, non‐receptor type 2 (PTPN2), endoplasmic reticulum aminopeptidase 1 (ERAP1), tumour necrosis factor receptor‐associated 3 interacting protein 2 (TRAF3IP2) and microRNA 146a (MIR146A) genes by allelic discrimination assays. Case‐control association studies and genotype/phenotype correlation analyses were performed. A higher risk to develop RA was observed for rs7574865 in the STAT‐4 gene, while the rs1800872 in the IL‐10 gene showed a protective effect. The presence of RF was associated significantly with rs1800872 variant in IL‐10, while rs2910164 in MIR146A was protective. ACPA were associated significantly with rs7574865 in STAT‐4. The SNP rs2233945 in the PSORS1C1 gene was protective regarding the presence of bone erosions, while rs2542151 in PTPN2 gene was associated with joint damage. Our results confirm that polymorphisms in STAT‐4 and IL‐10 genes confer susceptibility to RA. For the first time, we described that SNPs in PSORS1C1, PTPN2 and MIR146A genes were associated differently with a severe disease phenotype in terms of autoantibody status and radiographic damage in an Italian RA population.

Introduction

Rheumatoid arthritis (RA) is a chronic inflammatory autoimmune disease characterized by progressive joint destruction and influenced by both genetic and environmental factors 1. Epidemiological genetic data suggest that the heritability of RA is 53–60% 2. The relative risk for siblings of RA patients ranges from 7 to 12, suggesting that genetic factors contribute to disease susceptibility 3. During the past decade, several genes have been described as being associated with RA, although the major contribution to RA susceptibility has been identified in the human leucocyte antigen (HLA)‐DRB1 gene, referred to as the shared epitope 4. RA is a heterogeneous disease and patients may display different phenotypes that account for a variable prognosis. These prognostic factors include the presence or absence of rheumatoid factor (RF), anti‐citrullinated peptide antibodies (ACPA) and bone erosions 5. Several genes have been proved to contribute to the disease susceptibility and they may be associated with a particular pattern of disease and with the response or toxicity to therapy 6. Recently, a number of genetic polymorphisms has been associated with structural damage and increased rate of joint destruction 7. Indeed, it is important to identify novel genomic biomarkers associated not only with disease susceptibility, but also able to detect early those patients with negative prognostic factors who may benefit from a more aggressive therapeutic approach.

At first, we aimed to confirm the association described between RA and genes known for their susceptibility for the disease as signal transducer and activator of transcription‐4 (STAT‐4). Indeed, STAT‐4 seems to be one of the genes associated most with RA in different populations, and plays a key role in both T helper type 1 (Th1) and 17 (Th17) pathways 8, 9. This was useful, as we could confirm that our sample was representative of a Caucasian population and the sample size was powerful. We then explored genetic variants that have been investigated previously in RA with contradictory results such as the interleukin (IL)‐10 gene. This gene codes for IL‐10, an intriguing cytokine with pleiotropic effects that has anti‐inflammatory property, stimulates B cell proliferation, differentiation and survival 10 and promotes antibody isotype switching 11. Genetic variants in certain ethnic groups may over‐produce or lower IL‐10 production, with potential effects on the disease 12. Association with genes, such as tumour necrosis factor receptor‐associated 3 interacting protein 2 (TRAF3IP2), psoriasis susceptibility 1 candidate 1 (PSORS1C1) and protein tyrosine phosphatase, non‐receptor type 2 (PTPN2), was investigated in this study as they are involved in either Th17‐ or Th1‐mediated immune responses that are typical of RA 13, 14, 15. Increased expression of microRNA146a (MIR146a) has been documented in synovial fluid, synovial tissue, synoviocytes and peripheral blood mononuclear cells of RA patients. There is some evidence that MIR146A may inhibit Th1 response and enhance the activity of T‐regulatory cells in RA 16. New genes such as endoplasmic reticulum aminopeptidase 1 (ERAP1) were also studied as being related to other chronic inflammatory diseases, such as psoriatic arthritis 17.

Thus, the aim of our study was to investigate genomic variability in known and new genes as being associated with RA or with other autoimmune inflammatory diseases in order to verify and/or confirm their association with both disease susceptibility and clinical profile in a cohort of Italian RA patients, naive for biologic treatment. For this purpose, we performed a case–control association study, analysing selected polymorphisms (single nucletotide polymorphisms: SNPs) in STAT‐4, IL‐10, TRAF3IP2, PSORS1C1, MIR146A, PTPN2 and ERAP1 genes. We then performed a genotype–phenotype correlation analysis in order to evaluate any possible association with specific clinical phenotypes characterized by the presence of negative prognostic factors, such as RF, ACPA and bone erosions.

Materials and methods

The study included 192 RA patients and 278 age‐ and sex‐matched healthy controls (HC). All the patients and HC were Italian Caucasians. The study protocol was approved by the local ethics committee of the University of Rome ‘Tor Vergata’ (Italy). Written informed consent was obtained from patients and HC. Medical records of patients referred to the Rheumatology Outpatient Clinic at the Department of ‘Medicina dei Sistemi’, University of Rome ‘Tor Vergata’ (Italy), were analysed retrospectively. Patients were included into the study if they fulfilled the 2010 European League Against Rheumatism/American College of Rheumatology (EULAR/ACR) revised criteria for RA 18, they were at least 18 years of age and were naive for biological treatment. Clinical evaluation included the 28 tender and swollen joint count for disease activity score [DAS28, erythrocyte sedimentation rate (ESR)‐based], while laboratory assessment included ESR, C‐reactive protein (CRP), RF and ACPA. CRP and RF levels were tested by nephelometric measurement (normal range = 0–5 mg/l and 0–10 IU, respectively). ACPA were detected with a commercial third‐generation automated chemiluminescent kit: values > 20 IU were considered positive. The presence of bone erosions at diagnosis was evaluated in radiographs of hands and feet by the simple erosion narrowing score method by one experienced reader, who was blinded to clinical, biological and genetic data 19. Peripheral blood samples were obtained at the time of the enrolment from all included RA patients and controls in order to perform the genetic analyses. Samples were stored at −80°C until they were analysed. All patients were naive for biological treatments at the time of blood sampling.

DNA extraction and genotyping

Genomic DNA was isolated from peripheral blood mononuclear cells using a Qiagen blood DNA mini kit (Qiagen, Valencia, VA, USA). A panel of SNPs was selected on the basis of literature data in seven genes involved in immune response and inflammation already described as being associated with any autoimmune disease. The selected SNPs were the following: rs7574865 (STAT‐4); rs1800872 and rs3024505 (IL‐10); rs33980500, rs13190932 and rs13196377 (TRAF3IP2); rs2233945 (PSORS1C1); rs2910164 (MIR146A); rs2542151 and rs7234029 (PTPN2); and rs30187 and rs27524 (ERAP1). Genotyping was performed by allelic discrimination assay by TaqMan assays (Applied Biosystems, Foster City, CA, USA) and ABI PRISM 7500. Each assay was run with positive (samples confirmed previously by direct sequencing as heterozygous and/or variant homozygous) and negative controls.

Statistical analysis

The Hardy–Weinberg equilibrium was verified for all SNPs by the Pearson χ² test. Differences in allele and genotype frequencies between cases and controls were evaluated by Pearson's χ² test or by Fisher's exact test, when required. Odds ratios (ORs) with 95% confidence interval (CI) were calculated. A genotype–phenotype correlation analysis was performed in order to evaluate a possible correlation between the genetic variants and the presence of RF, ACPA and bone erosions. All statistical analyses were performed by SPSS version 13.0 (SSPS Inc., Chicago, IL, USA). Two‐tailed P‐values less than 0·05 were considered statistically significant.

Results

Clinical and demographic features of RA patients are reported in Table 1. All enrolled RA patients showed long‐standing disease. Negative prognostic factors were identified and were distributed as follows: 67·7% (n = 130) of cases tested positive for RF, ACPA were present in 71·7% (n = 137) and bone erosions were detected in 60% (n = 69) of patients.

Table 1.

Clinical data from the patient population

	RA patients (n = 192)
Age (years)	54·1 ± 13·2
Female (n/%)	147/76·5
Disease duration (years)	7·8 ± 9
Smoking (n/%)	33/49·3
Never smoking	34/50·7
Not known	125
CRP (mg/l)	11·5 ± 16·5
DAS28	5·2 ± 1·3
Rheumatoid factor‐positive (n/%)	130/67·7
ACPA‐positive (n/%)	137/71·7
Bone erosions at diagnosis (n/%)	69/60
No erosions	46/40
DMARDs (n/%)	148/77·1

Data are expressed as mean and standard deviation (s.d.) or absolute number and percentage. RA = rheumatoid arthritis; CRP = C‐reactive protein; DAS28 = 28 joint count disease activity score; ACPA = anti‐citrullinated protein antibody; DMARDs = disease modifying anti‐rheumatic drugs.

Case–control association analysis

Deviations from Hardy–Weinberg equilibrium for the investigated SNPs were not observed. Table 2 shows the genotype and allele distribution in HC and RA patients. As shown, we observed a significant association between the variant alleles of rs1800872 in IL‐10 gene and RA susceptibility. Subjects carrying the variant alleles (in heterozygous and homozygous status) showed a lower risk of developing RA in comparison with wild‐type individuals (OR = 0·63, P = 0·014). Moreover, we observed a higher risk of developing RA in subjects carrying the variant allele of rs7574865 in the STAT‐4 gene. Such association reached the statistical significance at allelic level (P = 0·035, with OR = 1·41), even if it was borderline at genotypical level (OR = 1·43; P = 0·06). Concerning the rs30187 in ERAP1 gene, we observed a trend for an association despite statistical significance not being reached.3

Table 2.

Case–control association analysis

				Genotypes*					Alleles
Gene	SNPs		n	wt	hz	var	P †	OR	wt	var	P	OR
								(95% CI)				(95% CI)
TRAF3IP2	rs33980500 C>T	Cases	191	161	29	1	0·54	1·18	351	31	0·46	1·20
		Controls	278	240	38	0		(0·70–1·98)	518	38		(0·74–1·97)
	rs13190932 G>A	Cases	191	168	22	1	0·57	1·19	358	24	0·47	1·23
		Controls	271	243	28	0		(0·66–2·13)	514	28		(0·70–2·16)
	rs13196377 G>A	Cases	190	167	22	1	0·32	1·35	356	24	0·27	1·38
		Controls	269	244	25	0		(0·74–2·46)	513	25		(0·78–2·46)
IL‐10	rs1800872 G>T	Cases	192	112	63	17	0·014	0·63	287	97	0·058	0·75
		Controls	278	130	124	24		(0·43–0·91)	384	172		(0·56–1·01)
	rs3024505 G>A	Cases	192	137	52	3	0·14	1·38	326	58	0·15	1·32
		Controls	270	209	58	3		(0·9–2·1)	476	64		(0·9–1·94)
STAT‐4	rs7574865 G>T	Cases	191	101	79	11	0·06	1·43	281	101	0·035	1·41
		Controls	243	150	87	6		(0·98–2·11)	387	99		(1·02–1·93)
MIR146A	rs2910164 G>C	Cases	192	109	69	14	0·41	0·86	287	97	0·47	0·90
		Controls	298	158	117	23		(0·60–1·24)	433	163		(0·67–1·20)
PSORS1C1	rs2233945 C>A	Cases	191	122	63	6	0·29	1·23	307	75	0·40	1·15
		Controls	289	198	81	10		(0·84–1·81)	477	101		(0·83–1·61)
ERAP1	rs30187 C>T	Cases	192	61	98	33	0·07	1·44	220	164	0·11	1·24
		Controls	256	103	114	39		(0·97–2·14)	320	192		(0·95–1·63)
	rs27524 C>T	Cases	192	63	95	34	0·25	1·26	221	163	0·23	1·18
		Controls	255	97	120	38		(0·85–1·86)	314	196		(0·90–1·55)
PTPN2	rs2542151 T>G	Cases	147	117	28	2	0·33	0·75	262	32	0·35	0·78
		Controls	122	91	29	2		(0·42–1·33)	211	33		(0·46–1·31)
	rs7234029 A>G	Cases	145	101	40	4	0·28	1·35	242	48	0·46	1·20
		Controls	123	93	25	5		(0·78–2·32)	211	35		(0·74–1·92)

SNP = single nucleotide polymorphism. Significant P‐values are reported in bold type. *wt = the homozygous genotype for the wild‐type allele; hz = the heterozygous genotype; var = the homozygous genotype for the variant allele. ^†Heterozygotes and variant homozygotes were considered together (1 df) in the comparisons between genotypes by χ² test. IL = interleukin; STAT‐4 = signal transducer and activator of transcription 4; PSORS1C1 = psoriasis susceptibility 1 candidate 1; PTPN2 = protein tyrosine phosphatase, non‐receptor type 2; ERAP1 = endoplasmic reticulum aminopeptidase 1; TRAF31P2 = tumour necrosis factor receptor‐associated 3 interacting protein 2; MIR146A = microRNA146a; OR = odds ratio; CI = confidence interval.

Table 3.

Genotype–phenotype association analyses

(a) Association with RF
Gene	SNP		Number of patients	Genotypes*			P **	OR	Alleles		P	OR
								(95% CI)				(95% CI)
IL‐10	rs1800872 G>T	RF‐positive	130	69	49	12	0·032	2·0	187	73	0·07	1·63
		RF‐negative	62	43	14	5		(1·06–3·8)	100	24		(0·97–2·74)
MIR146A	rs2910164 G>C	RF‐positive	130	80	42	8	0·05	0·55	202	58	0·05	0·63
		RF‐negative	62	29	27	6		(0·3–1·02)	85	39		(0·39–1·0)

(b) Association with ACPA
Gene	SNP		Number of patients	Genotypes*			P **	OR	Alleles		P	OR
								(95%CI)				(95% CI)
STAT‐4	rs7574865 G>T	ACPA‐positive	135	64	64	9	0·004	2·64	192	82	0·01	2·09
		ACPA‐negative	53	37	14	2		(1·34–5·18)	88	18		(1·18–3·69)

(c) Association with erosions presence
Gene	SNP		Number of patients	Genotypes*			P **	OR	Alleles		P	OR
								(95%CI)				(95% CI)
PSORS1C1	rs2233945 C>A	Erosions presence	69	51	16	2	0·042	0·44	118	20	0·06	0·52
		Erosions absence	45	25	18	2		(0·20–0·98)	68	22		(0·27–1·03)
PTPN2	rs2542151 T>G	Erosions presence	54	38	14	2	0·044	3·26	90	18	0·03	3·3
		Erosions absence	35	31	4	0		(1–10·77)	66	4		(1·07–10·21)

RF = rheumatoid factor; ACPA = anti‐citrullinated protein antibody; SNP = single nucleotide polymorphisms; OR = odds ratio; CI = confidence interval; STAT‐4 = signal transducer and activator of transcription 4; PSORS1C1 = psoriasis susceptibility 1 candidate 1; PTPN2 = protein tyrosine phosphatase, non‐receptor type 2; MIR146A = microRNA146a. Significant P‐values are reported in bold type. *wt = the homozygous genotype for the wild‐type allele; hz = the heterozygous genotype; var = the homozygous genotype for the variant allele. **Heterozygotes and variant homozygotes were considered together (1 df) in the comparisons between genotypes by χ² test.

Phenotype–genotype correlation analysis

We performed a phenotype–genotype correlation analysis considering the genotypes in relation to clinical characteristics of the RA patients. Interestingly, polymorphisms in IL‐10 and STAT‐4 genes were associated with specific clinical phenotypes. Indeed, the variant T allele of rs1800872 SNP in IL‐10, despite being a protective variant respect to RA disease development, was more present in RF‐positive than RF‐negative patients (OR = 2 and P = 0·032). Furthermore, the variant allele of rs7574865 SNP in the STAT‐4 gene was significantly more present in ACPA‐positive than in ACPA‐negative patients (OR = 2·64 and P = 0·004). Some polymorphisms, that were not associated with the disease, seemed to be associated with clinical phenotypes. The rs2910164 SNP in MIR146A gene were associated with RF; specifically, the variant C allele was more present in RF‐negative patients (OR = 0·55 with P = 0·05). Two polymorphisms seemed to be associated with the presence of bone erosions. Indeed, the rs2233945 SNP in PSORS1C1 gene and the rs2542151 SNP in PTPN2 gene seemed to be involved in the development of bone erosions such that the variant allele of rs2233945 was more present in patients who had no bone erosions (OR = 0·44 and P = 0·042). On the contrary, regarding the SNP in PTPN2 gene, the variant allele was more present in patients who had erosive joint damage with a higher risk compared with wild‐type patients (OR = 3·26 and P = 0·044).

Discussion

RA is a multi‐factorial chronic disease associated with the presence of RF and consists of clinical subsets of ACPA‐positive and ACPA‐negative patients, reflecting a different genetic predisposition and the involvement of a wide spectrum of mediators of both the innate and adaptive immune systems 20, 21, 22. The disease course is characterized by different phases, including a preclinical stage in which the identification of genomic biomarkers predictive of disease development and prognosis could prove extremely useful.

In the present study, we performed a retrospective analysis on a cohort of Italian RA patients, naive to biological treatment, in order to investigate the role of several genetic variants on disease susceptibility and phenotype. We investigated 12 SNPs in seven genes; some of them were already described as being associated with RA, while others were associated with different inflammatory diseases 23, 24.

Data from the literature suggest that STAT‐4, IL‐10 and PTPN2 polymorphisms may play a role in RA susceptibility 15, 25, 26, 27, as well as in other autoimmune diseases 28, 29. PSORS1C1 and ERAP1 genes were described previously as being associated with psoriasis 17, 29, while TRAF3IP2 polymorphisms were found to be associated with psoriasis, psoriatic arthritis and systemic lupus erythematous 14, 30, 31. Finally, MIR146A polymorphisms seem to contribute to systemic lupus erythematous and ankylosing spondylitis (AS) development 32, 33. In our case–control association analysis, we confirm that STAT‐4 and IL‐10 polymorphisms are associated with RA susceptibility. In particular, we described a higher risk of developing RA conferred by the rs7574865 in STAT‐4 gene, while rs1800872 in the IL‐10 gene showed a protective effect. The association with the STAT‐4 gene is consistent with previous studies 24, 26. The most recent meta‐analyses suggest that the rs7574865 confers a susceptibility that is independent from the presence of ACPA or RF 34. None the less, contradictory results are present in the literature regarding the association between SNPs in IL‐10 gene and RA susceptibility and severity, explained partly by the heterogeneity of the studies in terms of ethnicity and environment 35, 36, 37. In particular, the SNP rs1800872 seems to confer susceptibility to RA in Asians 35, while it seems to be protective in our Caucasian population. Further studies should address this issue, as polymorphisms of IL‐10 may alter transcription factors affecting IL‐10 production and the disease itself. Patients carrying putative IL‐10 low‐producer genotypes showed a reduced survival and activation of autoreactive B cells and a favourable response to anti‐tumour necrosis factor (TNF) treatment 12. For the first time, we have observed a trend for an association of rs30187 in ERAP1 gene and RA susceptibility. The role of this gene has been detailed in association with AS and with psoriasis 17. This gene encodes for a zinc metallopeptidase that has a role in processing for major histocompatibility complex class I presentation, cytokine receptor shedding, angiogenesis/blood pressure regulation and activation of macrophages. In our cohort, the relatively small size of the RA sample might affect the lack of a significant association with rs30187 in ERAP1 gene; therefore, this SNP should be investigated further in larger cohorts of RA patients.

We further described a possible predictive value of some gene variants on disease phenotype. We found that the studied IL‐10 and MIR146A polymorphisms were associated with RF status. Indeed, the T allele of rs1800872 in IL‐10 was associated with RF production 38. The C allele of rs2910164 in MIR146A shows a protective contribution. Polymorphisms in miRNA may alter miRNA expression and have been implicated in the pathogenesis of RA. MicroRNA (MIR or miRNA) are small non‐coding RNA that regulate gene expression at the post‐transcriptional level and serve as potential mediators and markers of inflammatory diseases. A variety of miRNA appear to be dysregulated in RA 39. Previous studies showed that mir‐146a was over‐expressed in both the serum and joints of RA patients 16. In our study, the variant allele of the MIR146A was associated with negative RF status, consistent with the evidence that this polymorphism leads to a reduced expression of mature mir‐146a 40, 41, 42, 43. Finally, it has also been suggested that the role of the rs2910164 polymorphism may depend upon gender, thus larger cohorts are required to clarify this issue 44. The STAT‐4 variant was associated highly with the presence of ACPA. Indeed, in our population, the T allele conferred a more than double risk to develop ACPA. As mentioned above, a subgroup analysis according to the absence or presence of RF and ACPA revealed that the association between the STAT‐4 rs7574865 polymorphism and RA might be independent of their presence 45. However, variants in STAT genes might contribute differently to susceptibility to RA in seropositive and seronegative patients 46. In another study, patients with early arthritis who are homozygous for the T allele of rs7574865 in STAT‐4 developed a more severe disease form with higher disease activity and disability 47. The most intriguing associations were observed between rs2233945 in PSORS1C1 or rs2542151 in PTPN2 and the risk of developing bone erosions. To our knowledge, there are no data in the literature regarding the PSORS1C1 and PTPN2 genes in relation to bone erosions. PSORS1C1 is a susceptibility gene for psoriasis and psoriatic arthritis 48, 49. Only one study, to date, has reported several PSORS1C1 SNPs, different from rs2233945, as predictive of RA susceptibility 13. In our study, we did not observe any association with RA susceptibility, but the variant allele of rs2233945 seems to be a protective factor versus radiographically demonstrated bone damage. Of interest, a recent study revealed that the inhibition of PSORS1C1 expression in synovial RA fibroblasts cultured in vitro was associated with the reduction of IL‐17 and IL‐1β levels and a decrease of cell proliferation 13. IL‐17 is a well‐known osteoclastogenic factor that has been found implicated in bone erosion development. Thus, PSORS1C1 might protect from bone damage in RA by inhibiting IL‐17 production from synovial fibroblasts. Regarding PTPN2, some SNPs were reported previously to be associated with RA in populations of both Japanese and European ancestry 50, 51. A recent study demonstrated that PTPN2 was over‐expressed in synovial fibroblasts from RA patients and regulated IL‐6 production, cell death and autophagy 15. This could be consistent with our observation that the rs2542151 variant allele gives a higher risk to develop joint erosion (OR > 3), although we were not able to detect any association with disease susceptibility in our cohort. These data, if confirmed in larger cohorts of patients with inflammatory arthritis, could be crucial, as some patients with undifferentiated non‐erosive arthritis and genetic risk factors for erosions may benefit of an aggressive treatment aimed at preventing bone damage, while others carrying protective risk factors may undergo a different tight control strategy.

In conclusion, our study confirms the association of STAT‐4 and IL‐10 polymorphisms with RA and with autoantibody status. We describe for the first time an association between MIR146A and RF production and between PSORS1C1 and PTPN2 SNPs and radiographic bone damage. These results need to be confirmed in larger cohorts of RA patients. Functional studies are also required to clarify the role of these SNPs in the pathogenic mechanisms of the disease.

Author contributions

C. C., C. P. and P. B. conceived and designed the experiments; C. C., P. C. and C. P. wrote the manuscript; S. R. and C. P. performed the genetic analysis; C. C. and S. R. performed the statistical analysis; P. C., P. T. and R. P. enrolled patients and collected samples; G. N., R. P. and P. B. supervised the research. All authors revised and approved the final manuscript.

Disclosure

The authors declare no commercial or financial conflicts of interest.