STAT4 is a genetic risk factor for systemic sclerosis having additive effects with IRF5 on disease susceptibility and related pulmonary fibrosis

Authors


Abstract

Objective

Systemic sclerosis (SSc) belongs to the group of connective tissue disorders (CTDs), among which are several disorders characterized by a type I interferon (IFN) signature. The recent identification of an association between IRF5 and SSc further highlights a key role for IFN. STAT4, which encodes STAT-4, contributes to IFN signaling, and its genetic variants were found to be associated with CTDs. The aim of this study was to determine whether the STAT4 rs7574865 single-nucleotide polymorphism is associated with SSc, and whether it interacts with IRF5.

Methods

Both the STAT4 rs7574865 and IRF5 rs2004640 polymorphisms were genotyped in 1,855 individuals of French Caucasian origin comprising a discovery set of 440 patients with SSc and 485 control subjects and a replication set of 445 patients with SSc and an additional 485 control subjects.

Results

STAT4 rs7574865 was shown to be associated with SSc (P = 0.001, odds ratio [OR] 1.29, 95% confidence interval [95% CI] 1.11–1.51). This association was not restricted to a particular phenotype. An additive effect of the STAT4 rs7574865 T allele and the IRF5 rs2004640 T allele was observed, resulting in a multiplicatively increased 1.28-fold risk of SSc. The OR for SSc was 2.72 (95% CI 1.86–3.99) for combinations of genotypes with ≥3 risk alleles. An additive effect was also detected for fibrosing alveolitis: carriage of at least 3 risk alleles appeared to be an independent risk factor (P = 2.2 × 10−4, OR 1.97, 95% CI 1.28–3.04).

Conclusion

Our results establish STAT4 rs7574865 as a new SSc genetic susceptibility factor. STAT4 and IRF5 act with additive effects in terms of susceptibility to both SSc and SSc-related fibrosing alveolitis.

Various association studies have convincingly identified STAT4 as a susceptibility factor for systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), Sjögren's syndrome (SS), inflammatory bowel disease, and type 1 diabetes mellitus (1–6). STAT4 encodes for STAT-4, which transduces interleukin-12 (IL-12), IL-23, and type I interferon (IFN) cytokine signals in T cells and monocytes (7). The STAT4 susceptibility variant is the single-nucleotide polymorphism (SNP) rs7574865, although its functional consequence remains unclear. SNPs identified to be in strong linkage disequilibrium with rs7574865 are located in the third intron of STAT4, suggesting the involvement of a splice variation or regulatory effects (8).

Systemic sclerosis (SSc) is a connective tissue disease (CTD) characterized by early generalized microangiopathy, immune system disturbances, and massive deposits of extracellular matrix in the connective tissue. The higher IFNα induction in sera from patients with diffuse cutaneous SSc (dcSSc) compared with that in sera from patients with limited cutaneous SSc (lcSSc) as well as that in sera from patients with lung fibrosis suggests that IFNα may contribute to the fibrotic process (9). Thus, taking into account that immunogenetic studies have recently provided major results in the field of CTDs, that innate immunity and the type I IFN system play a pivotal role in these different autoimmune diseases (10), and the recently reported involvement of IRF5 rs2004640 in the genetic background of SSc (11), STAT4 was thus considered as a candidate gene. In the present study, we investigated whether STAT4 rs7574865 is associated with SSc in a French Caucasian population and also investigated its putative gene–gene interaction with IRF5.

PATIENTS AND METHODS

Study population and study design

We performed a large case–control association study that included a replication strategy, as previously described (11). The discovery set included 440 patients with SSc and 485 control subjects from the French network of Ile-de-France hospitals. The replication set, which included the participation of French centers from outside the first geographic area, included 445 patients with SSc and 485 control subjects. All of the patients with SSc were classified according to cutaneous subtypes described by LeRoy et al (12).

The 2 control groups comprised healthy, unrelated French blood donors or volunteers who were ethnically matched to the patients with SSc. All study subjects were of French Caucasian ethnicity, which was defined as all 4 grandparents being French Caucasian. None of the control subjects had a personal or family history of autoimmune disease, according to the entry criteria previously reported by the Multiple Autoimmune Disease Genetics Consortium (13). The ethics committee of Cochin Hospital approved the study, and all subjects provided written informed consent. Among patients with SSc, those identified as having the co-occurrence of a known STAT4 rs7574865–associated autoimmune disease (SLE, RA, SS, type 1 diabetes mellitus) were excluded from the analyses. The characteristics of the study subjects are shown in Table 1.

Table 1. Characteristics of the 801 SSc patients with available STAT4 rs7574865 genotype information*
CharacteristicDiscovery set (n = 402)Replication set (n = 399)Combined SSc samples (n = 801)
  • *

    Except where indicated otherwise, values are the number/number of patients for whom data were available (%). Patients with co-occurrence of a STAT4 rs7574865 T allele–associated autoimmune disease were excluded from the analyses. SSc = systemic sclerosis; ANA = antinuclear antibody; anti–topo I = anti–topoisomerase I antibody; ACA = anticentromere antibody; CT = computed tomography.

Female sex, %83.982.383.1
Age, mean ± SD years57.7 ± 13.257.3 ± 12.357.5 ± 12.8
Disease duration, mean ± SD years11.8 ± 7.911.2 ± 9.211.5 ± 8.6
Limited cutaneous SSc241/374 (64.4)249/386 (64.5)490/760 (64.5)
Diffuse cutaneous SSc133/374 (35.6)137/386 (35.5)270/760 (35.5)
ANA positivity312/361 (86.4)335/372 (90.5)647/733 (88.3)
Anti–topo I positivity103/353 (29.2)95/372 (25.5)198/725 (27.3)
ACA positivity118/353 (33.4)176/372 (47.3)294/725 (40.5)
Fibrosing alveolitis on CT scan179/374 (47.8)137/374 (36.6)316/748 (42.2)

Genotyping

The STAT4 rs7574865 and IRF5 rs2004640 SNPs were genotyped using a competitive allele-specific polymerase chain reaction system (KASpar genotyping; an improved fluorescence resonance energy transfer method developed by KBiosciences, Hoddeston, UK [http://www.kbioscience.co.uk/]), as previously described (11). The average genotype completeness was 98% for the SSc and control samples. The accuracy was >99%, according to duplicate genotyping of 10% of all samples, using the TaqMan SNP genotyping assay allelic discrimination method (Applied Biosystems, Foster City, CA).

Statistical analysis

In case an association was detected in both the discovery and replication sets, analyses of particular phenotypic subgroups were planned in the combined populations in order to increase the power of detection.

The statistical analyses were performed using the R computer software package (version 2.6.0; Free Software Foundation, Boston, MA). The level of significance for all the tests was set to a Type I error rate of α = 5%. The Bonferroni correction was applied for all the tests performed for a “generating hypothesis step” when comparing patients with SSc and control subjects (10 phenotypic subsets). P values remaining significant after the adjustment for multiple testing are termed corrected P (Pcorr). Odds ratios (ORs) and their 95% confidence intervals (95% CIs) are reported. Conformity with Hardy-Weinberg equilibrium was tested using a classic chi-square test with 1 degree of freedom in control subjects.

Individual association analyses of STAT4 rs7574865 with SSc were performed by comparing genotypes in patients and control subjects, using Fisher's exact test. The corresponding ORs were assessed using a standard logistic regression analysis, with the most frequent genotype taken as the reference (i.e., the STAT4 rs7574865 GG genotype). A logistic regression model was used to estimate gene–gene interactions between STAT4 rs7574865 and IRF5 rs2004640 and to determine the additive effects of these 2 SNPs. This analysis was performed taking into account the combination of both genotypes, leading for one given individual to a risk allele count ranging from 0 to 4. ORs were computed using the logistic regression model, using 0 risk alleles as the reference. Fisher's exact test was used to test for the difference in STAT4 and IRF5 risk allele counts between patients and control subjects.

Power calculations were driven through a standard noncentral chi-square approach, as previously described (14). The discovery set of 440 patients with SSc and 485 control subjects provides power of >82% to detect an association between SSc and the STAT4 T allele previously observed in combined SLE populations (4), with an OR of 1.55 at the 5% significance level.

RESULTS

Discovery set.

Among the 440 patients with SSc, 34 were excluded prior to analysis because of co-occurrence of a STAT4 rs7574865 T allele–associated disease. Genotype frequencies were in Hardy-Weinberg equilibrium in the control population. The case–control association study performed in the discovery set (402 patients with SSc and 481 control subjects [4 genotypes for control subjects were missing]) revealed an association between the STAT4 rs7574865 T allele (P = 0.039, OR 1.26, 95% CI 1.01–1.56), the homozygous TT genotype (P = 0.017, OR 2.03, 95% CI 1.12–3.65), and SSc (Table 2).

Table 2. Genotype and allele distribution of the STAT4 rs7574865 polymorphism in patients with SSc and control subjects*
 TT, no. (%)GT, no. (%)GG, no. (%)T, %P vs. controlsOR (95% CI)
TTTGTTT/GTTTTGTTT/GT
  • *

    P values were determined by chi-square test. The GG genotype was used as the reference, with the value set at 1. SSc = systemic sclerosis; OR = odds ratio; 95% CI = 95% confidence interval.

Discovery set            
 SSc (n = 402)31 (7.7)156 (38.8)215 (53.5)27.10.0390.0170.380.141.26 (1.01–1.56)2.03 (1.12–3.65)1.13 (0.86–4.19)1.22 (0.94–1.59)
 Controls (n = 481)20 (4.2)180 (37.4)281 (58.4)22.9        
Replication set            
 SSc (n = 399)24 (6.0)164 (41.1)211 (52.9)26.60.00990.150.0110.00691.35 (1.07–1.66)1.55 (0.84–2.83)1.44 (1.09–1.89)1.45 (1.11–1.89)
 Controls (n = 483)22 (4.5)162 (33.5)299 (61.9)21.3        
Combined samples            
 All SSc (n = 801)55 (6.9)320 (39.9)426 (53.2)26.80.00100.00640.0160.00311.29 (1.11–1.51)1.78 (1.17–2.72)1.27 (1.05–1.55)1.33 (1.10–1.61)
 All controls (n = 964)42 (4.3)342 (35.5)580 (60.2)22.1        

Replication set.

Prior to the analysis, 38 of the 445 patients with SSc were excluded because of co-occurrence of a STAT4 rs7574865 T allele–associated disease. Genotype frequencies in the control population were in Hardy-Weinberg equilibrium. An association was observed between the STAT4 rs7574865 T allele (P = 0.0099, OR 1.35, 95% CI 1.07–1.66), the GT heterozygous genotype (P = 0.011, OR 1.44, 95% CI 1.09–1.89), and SSc (n = 399 [8 genotypes were missing]) (Table 2).

Combined populations.

When the analysis was performed in the pooled samples, the STAT4 rs7574865 T allele was observed on 26.8% of chromosomes in patients with SSc compared with 22.1% of chromosomes in control subjects (P = 0.0010, OR 1.29, 95% CI 1.11–1.51) (Table 2). The STAT4 rs7574865 TT, GT, and GT/TT genotypes were shown to be associated with SSc (Table 2).

STAT4 rs7574865 variant and SSc phenotypes.

STAT4 rs7574865 and autoantibody status.

The STAT4 rs7574865 T allele was observed on 26.9% of chromosomes in antinuclear antibody (ANA)–positive patients with SSc compared with 22.1% of chromosomes in control subjects (Pcorr = 0.01, OR 1.30, 95% CI 1.11–1.53) (Table 3). The association of the STAT4 rs7574865 T allele with SSc was independent of the subtype of ANA (data not shown). Association between the homozygous TT genotype and SSc was restricted to ANA-positive SSc (Pcorr = 0.02, OR 1.95, 95% CI 1.26–3.01). No association of the TT, GT, and GT/TT genotypes was detected, regardless of both anticentromere antibody (ACA) and anti−topoisomerase I (anti–topo I) status.

Table 3. Genotype and allele distribution of the STAT4 rs7574865 polymorphism in patients with SSc and control subjects, according to the phenotype of the patients*
 TT, no. (%)GT, no. (%)GG, no. (%)T, %P vs. controlsPcorr vs. controlsOR (95% CI)
TTTGTTT/GTTTTGTTT/GTTTTGTTT/GT
  • *

    The GG genotype was used as the reference, with the value set at 1. P values for comparisons of T allele and TT genotype frequencies among patients with diffuse cutaneous systemic sclerosis (SSc) versus patients with limited cutaneous SSc and among patients with fibrosing alveolitis versus patients without fibrosing alveolitis were as follows: for the T allele, 0.93 and 0.19, respectively; for the TT genotype, 0.83 and 0.18, respectively. OR = odds ratio; 95% CI = 95% confidence interval; ANA = antinuclear antibody; NS = not significant.

  • By chi-square test or Fisher's exact test.

  • Corrected P (Pcorr) values were less than 0.05 after Bonferroni correction for multiple comparisons.

SSc                
 ANA positive (n = 647)49 (7.6)251 (38.8)347 (53.6)26.90.00150.00210.0560.00930.010.02NSNS1.30 (1.11–1.53)1.95 (1.26–3.01)1.23 (0.99–1.51)1.31 (1.07–1.59)
 Diffuse cutaneous (n = 270)20 (7.4)106 (39.3)144 (53.3)27.00.01610.02140.130.0439NSNSNSNS1.31 (1.05–1.63)1.92 (1.09–3.37)1.25 (0.94–1.66)1.32 (1.01–1.73)
 Limited cutaneous (n = 490)34 (6.9)195 (39.8)261 (53.3)26.80.00450.01430.0460.01180.04NSNSNS1.29 (1.08–1.55)1.80 (1.12–2.89)1.27 (1.01–1.59)1.33 (1.06–1.65)
 Fibrosing alveolitis (n = 316)27 (8.5)127 (40.2)162 (51.3)28.60.00080.00120.0370.00540.0080.012NSNS1.42 (1.16–1.73)2.30 (1.38–3.85)1.33 (1.02–1.74)1.44 (1.11–1.85)
 No fibrosing alveolitis (n = 432)27 (6.2)167 (38.7)238 (55.1)25.60.04370.080.150.07NSNSNSNS1.21 (1.01–1.46)1.57 (0.94–2.60)1.19 (0.94–1.51)1.23 (0.98–1.55)
All controls (n = 964)42 (4.3)342 (35.5)580 (60.2)22.1            

STAT4 rs7574865 and fibrotic phenotype.

The frequencies of the STAT4 risk allele and the TT and TT/GT genotypes were increased in both patients with dcSSc and those with lcSSc when compared with control subjects (Table 3). We observed an increase in the frequency of the STAT4 rs7574865 T allele restricted to patients with lcSSc when compared with control subjects (26.8% versus 22.1%), and this difference reached statistical significance after correction for multiple comparisons (Pcorr = 0.04, OR 1.29, 95% CI 1.08–1.55). Conversely, no difference was observed when comparing the dcSSc and lcSSc subgroups (Table 3).

We observed a strong association between the STAT4 rs7574865 T allele and the presence of fibrosing alveolitis. The STAT4 T allele was observed on 28.6% of chromosomes in SSc patients with fibrosing alveolitis compared with 22.1% of control chromosomes (Pcorr = 0.008, OR 1.42, 95% CI 1.16–1.73). The association was stronger when TT genotypes were considered: the TT genotype was observed in 8.5% of SSc patients with fibrosing alveolitis compared with 4.3% of control subjects (Pcorr = 0.012, OR 2.30, 95% CI 1.38–3.85). In contrast, no association between STAT4 rs7574865 and SSc without pulmonary fibrosis was observed (Table 3). However, a comparison of the frequencies of both the STAT4 rs754865 T allele and the TT genotype between SSc patients with and those without fibrosing alveolitis did not reach statistical significance (Table 3).

IRF5 rs2004640 variant and SSc.

Regarding IRF5 rs2004640 in the global sample, we observed an association with SSc of a magnitude similar to that previously described by our group (11). The frequency of the rs2004640 T allele was 56.7% in patients with SSc compared with 50.2% in controls (P = 8.92 × 10−5, OR 1.31, 95% CI 1.14–1.49). The homozygous TT genotype was observed in 31.8% of patients with SSc compared with 24.9% of control subjects (P = 8.52 × 10−5, OR 1.72, 95% CI 1.31–2.25). The rs2004640 T allele was also shown to be strongly associated with SSc-related fibrosing alveolitis (P = 8.29 × 10−6, OR 1.52, 95% CI 1.26–1.82) (additional information may be obtained from the corresponding author).

Joint effects of the STAT4 rs7574865 and IRF5 rs2004640 risk alleles.

Joint effects on SSc.

We recently reported an association between IRF5 rs2004640 and SSc, particularly in the subgroup of SSc patients with fibrosing alveolitis (11). In the present study, we investigated the joint effect of the STAT4 rs7574865 and IRF5 rs2004640 T alleles on SSc susceptibility. The overall significance for the difference in risk allele counts between patients and control subjects was high, with a 1-sided P value of 1.37 × 10−7, using an additive logistic regression model. No evidence for dominance or interactions was observed between the 2 risk alleles using multiple logistic regression analysis, which ruled out an epistatic effect. We observed an additive effect of both risk alleles (the IRF5 rs2004640 T allele and the STAT4 rs7574865 T allele) on SSc susceptibility. The ORs for SSc were 1.72 (95% CI 1.23–2.14) for carriers of 1 risk allele, 2.17 (95% CI 1.55–3.04) for carriers of 2 risk alleles, and 2.72 (95% CI 1.86–3.99) for carriers of 3 or 4 risk alleles. Figure 1 shows the ORs for SSc patients with 1, 2, or ≥3 risk alleles, using individuals with no risk alleles as reference (additional information may be obtained from the corresponding author). As can be seen, the risk of SSc increases in a multiplicative manner as a function of the number of risk alleles, with a 1.28-fold increase in the OR for each additional risk allele.

Figure 1.

Joint effects of the STAT4 and IRF5 risk alleles on systemic sclerosis (SSc) susceptibility. The results of a linear regression analysis show a multiplicative effect of the T allele of IRF5 rs2004640 and the T allele of STAT4 rs7574865 on SSc susceptibility. The odds ratios with 95% confidence intervals are shown as a function of the number of risk alleles of SSc. The slope of the line corresponds to a 1.28-fold increase in the OR for each additional risk allele.

Joint effects on SSc-related fibrosing alveolitis.

Multiple logistic regression analysis showed a nonmultiplicative effect of both the IRF5 rs2004640 and STAT4 rs7574865 risk alleles, with a threshold of 3–4 risk alleles for an increasing risk of SSc. We observed a strong association between carriage of at least 3 risk alleles and SSc lung interstitial disease (P = 0.002, OR 1.786, 95% CI 1.235–2.582). The multivariate logistic regression model identified both the dcSSc subtype and anti–topo I antibodies as independent risk factors for SSc-related fibrosing alveolitis. Including the variable of at least 3 risk alleles in the multivariate model together with the dcSSc subtype and anti–topo I antibodies provided evidence of a remaining independent significant effect of the risk alleles (P = 2.2 × 10−4, OR 1.97, 95% CI 1.28–3.04) (additional information may be obtained from the corresponding author).

DISCUSSION

The results of this study show a role of STAT4 in the SSc genetic background and corroborate the notion that related autoimmune diseases share common risk variants and the notion of a shared common pathway among CTDs. Our case–control association study demonstrates a significant association between STAT4 rs7574865 and SSc. Our strategy, which included an independent replication step and the exclusion of SSc patients presenting with co-occurrence of another known STAT4 T allele–associated disease, supports the significance of our findings. The observed frequency of 22.1% for the risk allele in our control population was in good agreement with that previously reported (3, 4, 8).

In our French Caucasian population, individuals carrying 1 or 2 copies of the STAT4 rs7574865 T allele were at increased risk for the development of SSc. Analysis of the ORs in the combined data suggested a dose effect, since individuals carrying the homozygous TT genotype had a corresponding OR (1.78, 95% CI 1.17–2.72) not included in the confidence interval given by the GT genotype (OR 1.27, 95% CI 1.05–1.55).

Regarding the autoantibody status of the patients with SSc, the association of the STAT4 risk allele remained statistically significant only in the ANA-positive subgroup (Pcorr = 0.01, OR 1.30, 95% CI 1.11–1.53) without any specificity for ACAs or anti–topo I antibodies. To date, regarding other STAT4 rs7574865 T allele–associated diseases, the association between the risk allele and specific autoantibody production remains unclear. In contrast to the previously reported association between IRF5 rs2004640 and SSc, our results suggest that the involvement of STAT4 rs7574865 in SSc susceptibility is not restricted to a particular cutaneous subset of SSc. Regarding the fibrosing alveolitis phenotype, a strong association was detected between the homozygous TT genotype in SSc patients with fibrosing alveolitis compared with control subjects (OR 2.30, 95% CI 1.38–3.85). Nevertheless, the results of intracohort tests failed to identify a difference between SSc subsets. Therefore, larger sample sizes allowing further intracohort tests are required. Interestingly, a recent case–control association study performed in a European population identified STAT4 rs7574865 as a SSc genetic risk factor, but this association was restricted to the lcSSc subgroup (15).

Taking into account the recently reported additive effect of IRF5 and STAT4 in SLE susceptibility (8), we investigated the joint effect of both risk alleles in SSc susceptibility. We observed a multiplicative increase in the OR by a factor of 1.28 for each additional risk allele and a considerably increased risk of SSc for those 18.4% of individuals carrying at least 3 risk alleles. This point is of major interest, because an additive effect for a contribution to a general loss of tolerance has been reported in a Caucasian SLE population (8). Regarding the pulmonary fibrosis phenotype, logistic regression analysis revealed an increased risk of SSc-related fibrosing alveolitis in the presence of at least 3 risk alleles, and multivariate analysis demonstrated that carriage of at least 3 STAT4/IRF5 risk alleles is as an independent risk factor for SSc-related fibrosing alveolitis. Hence, both STAT4 and IRF5 could contribute to a disease-specific phenotype.

Identification of IRF5 and STAT4 as new SSc genetic susceptibility factors raises the question of the contribution of innate immunity in SSc pathogenesis and in the fibrotic process. Although the functional significance of association between the STAT4 rs7574865 variant and SSc remains unclear, a functional effect of a STAT4 SLE risk haplotype tagged by rs7574865 has been recently suggested (8).

The identification of STAT4 as a genetic susceptibility factor and its additive effects with IRF5 for various autoimmune diseases, at least SSc and SLE, is an important step toward understanding type I IFN–driven autoimmunity (10). However, to date, many other STAT4 and IRF5 polymorphisms have been identified as risk factors for other diseases and will have to be investigated in SSc. Furthermore, our results require replication in other populations.

In summary, our study shows a genetic association between STAT4 and SSc. We also demonstrate a strongly increased risk of SSc and fibrosing alveolitis in individuals carrying multiple IRF5 and STAT4 risk alleles. The additive effect observed between IRF5 and STAT4, particularly in the subgroup of patients with fibrosing alveolitis, suggests that combining information from common risk polymorphisms could improve disease prediction and may be useful for risk stratification. These data provide new insight into the pathogenesis of SSc, underlying the pivotal role of innate immunity and the type I IFN pathway.

AUTHOR CONTRIBUTIONS

All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Dieudé had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Dieudé, Allanore.

Acquisition of data. Wipff, Ruiz, Hachulla, Diot, Granel, Sibilia, Tiev, Mouthon, Cracowski, Carpentier, Amoura, Fajardy, Avouac, Meyer, Kahan, Allanore.

Analysis and interpretation of data. Dieudé, Guedj, Boileau, Allanore.

Acknowledgements

We thank Dr. Joëlle Benessiano, CRB, Hôpital Bichat, Paris, for providing and storing the DNA extracts for all control samples.

Ancillary