SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

Objective

Rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE) share some genetic factors such as HLA, PTPN22, STAT4, and 6q23. The aim of this study was to determine whether 9 other SLE genetic factors are also implicated in RA susceptibility.

Methods

A characteristic single-nucleotide polymorphism (SNP) in each of 9 genetic factors, ITGAM (rs1143679), C8orf13–BLK (rs13277113), TYK2 (rs2304256), 1q25.1 (rs10798269), PXK (rs6445975), KIAA1542 (rs4963128), MECP2 (rs17435), BANK1 (rs17266594), and LY9 (rs509749), was studied in 1,635 patients with RA and 1,906 control subjects from Spain. The rs7574865 SNP in STAT4 was also included. Analyses were conducted globally and after stratification by sex and clinical features (anti–cyclic citrullinated peptide and rheumatoid factor, shared epitope, rheumatoid nodules, radiographic changes, sicca syndrome, and pneumonitis).

Results

No association was observed between RA and any of the 9 newly identified SLE genetic factors. A meta-analysis using previous data was consistent with these results. In addition, there were no significant differences between individuals with and those without each of the clinical features analyzed, except the frequency of the minor allele in the C8orf13–BLK locus that was decreased in patients with sicca syndrome (14.6% versus 22.4% in controls; P = 0.003).

Conclusion

None of the 9 recently identified SLE risk factors showed association with RA. Therefore, common genetic factors affecting the pathogenesis of these 2 disorders seem to be limited, revealing that the genetic component contributes to the different expression of these diseases.

Rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE) share a complex etiology encompassing genetic, environmental, and stochastic components (1, 2). Loss of tolerance to self antigens, which leads to stimulation of lymphocytes and other immune cells, release of cytokines, activation of complement, and the production of autoantibodies, contributes to the pathogenesis of both RA and SLE (3, 4). RA and SLE also share genetic factors such as those in the HLA, PTPN22, STAT4, and 6q23 loci (5–8); however, their respective clinical phenotypes are clearly different. RA is characterized by symmetric erosive arthritis of the peripheral joints with pannus formation that is chronic and progressive, which causes a characteristic pattern of pain, rigidity, and deformities. In contrast, arthritis is only one among the diverse array of clinical manifestations of SLE and typically is not erosive, does not have a progressive course, and most often is not symmetric (9). Malar rash, photosensitivity, serositis, nephritis, peripheral and central nerve system disease, and cytopenia are indicative of SLE but not RA. The 2 clinical syndromes overlap only in some exceptional patients (a clinical situation that is referred to as rhupus by many investigators). Therefore, the 2 diseases should have differential etiologic factors. Such etiologic factors could be genetic, with some being specific for SLE and others being specific for RA. Alternatively, the differential factors could be environmental or stochastic.

A common genetic component of RA, SLE, and other autoimmune diseases has been hypothesized (10, 11). This hypothesis was triggered by the observation of an increased concurrence of several autoimmune diseases in members of the same family (12, 13) and was strongly reinforced by the discovery of the role of HLA alleles in most autoimmune diseases (14) and, later, by genetics studies in families showing that linkage to different autoimmune diseases clustered in overlapping loci (15). Similar findings were also observed in animal models of autoimmune diseases (16). More recently, some genetic factors with a shared effect in many of these diseases have been identified. The most generally shared is PTPN22, but other examples include STAT4, the 6q23 locus, and possibly CTLA4 and IRF5 (5–8, 17, 18). Now, thanks to genome-wide association studies, we are beginning to have more comprehensive knowledge of the genetic component of these diseases. The identification of more genetic factors will allow us to determine whether those that are shared by several diseases are the rule or the exception. As a consequence, we will make advances in the identification of differential disease factors.

At the time when this study was planned, the number of recently identified genetic factors in SLE was larger than the number of such factors in RA. Therefore, we examined the role in RA of a representative SNP for each of 10 recently identified SLE genetic factors: STAT4 (which is already known to be shared by SLE and RA), ITGAM, C8orf13–BLK, PXK, TYK2, KIAA1542, 1q25.1, BANK1, MECP2, and LY9 (6). All except LY9 have consistently been associated with SLE (19). However, only the SNP in STAT4 was associated with RA in our current comparison of more than 1,600 patients and 1,900 healthy control subjects. A meta-analysis including previous data led to similar results. Therefore, it seems that RA does not share most of the new SLE genetic factors, indicating that the genetic component is a significant contributor to the different expression of these 2 diseases.

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

Sample collection.

We used DNA samples from 1,635 patients with RA and 1,906 healthy control subjects of Spanish ancestry (281 of the control subjects were previously described by our group in a study of SLE susceptibility [19]). All patients with RA met the 1987 revised American College of Rheumatology (formerly, the American Rheumatism Association) classification criteria (20). The clinical characteristics of the patients are shown in Table 1. Patients and control subjects gave written informed consent to participate in the study. Sample collection was approved by the respective ethics committees.

Table 1. Clinical characteristics of the 1,635 patients with rheumatoid arthritis*
  • *

    Except where indicated otherwise, values are the percent. When data were not available for all patients, the number of patients for whom data were available is shown.

Female sex75.0
Age at disease onset, median (interquartile range) years48 (37–57)
Morning stiffness96.1
Arthritis in ≥3 joint areas99.7
Arthritis in hand joints99.0
Symmetric arthritis99.3
Rheumatoid nodules (n = 1,291)20.1
Rheumatoid factor (n = 1,504)72.7
Erosions (n = 1,502)70.4
Sicca syndrome (n = 903)14.8
Interstitial pneumonitis (n = 1,520)2.8
Shared epitope carrier (n = 594)54.7
Anti–citrullinated protein antibody positive (n = 655)66.1

SNP genotyping.

A strongly SLE-associated SNP for each of 10 recently identified SLE genetic factors was selected (Table 2). The 10 SNPs were amplified in a single polymerase chain reaction (PCR) (Qiagen Multiplex PCR Kit; Qiagen, Chatsworth, CA) with 20 ng of genomic DNA and 0.2 μM of each primer. PCR products were purified by digestion with exonuclease I (Epicentre Technologies, Madison, WI) and shrimp alkaline phosphatase (SAP; GE Healthcare, Barcelona, Spain). Purified PCR products were included in a single-base extension reaction with the SNaPshot Multiplex Kit (Applied Biosystems, Foster City, CA) and specific probes. After a second purification with SAP (GE Healthcare), samples were analyzed in the ABI Prism 3130xl Genetic Analyzer (Applied Biosystems), and genotypes were assigned using GeneMapper software (Applied Biosystems). (Primer and oligonucleotide sequences are available from the corresponding author.)

Table 2. Allele frequencies for the 10 SNPs identifying SLE genetic factors*
SNP (locus)Minor allele frequency, no. (%) 
RA patientsControl subjectsOR (95% CI)
  • *

    SLE = systemic lupus erythematosus; RA = rheumatoid arthritis; OR = odds ratio; 95% CI = 95% confidence interval.

  • P = 0.028.

  • This single-nucleotide polymorphism (SNP), which is in chromosome X, was assessed by analysis of allelic frequencies in women and by carrier analysis in men.

rs7574865 (STAT4)665/2,768 (24.0)784/3,612 (21.7)1.14 (1.0–1.3)
rs1143679 (ITGAM)561/3,264 (17.2)608/3,808 (16.0)1.09 (1.0–1.2)
rs13277113 (C8orf13–BLK)757/3,268 (23.2)852/3,810 (22.4)1.05 (0.9–1.2)
rs2304256 (TYK2)842/3,262 (25.8)1,025/3,804 (26.9)0.94 (0.8–1.0)
rs10798269 (1q25.1)1,042/3,268 (31.9)1,285/3,812 (33.7)0.92 (0.8–1.0)
rs4963128 (KIAA1542)1,013/3,264 (31.0)1,214/3,812 (31.8)0.96 (0.9–1.1)
rs6445975 (PXK)790/3,270 (24.2)851/3,814 (22.3)1.11 (1.0–1.2)
rs17266594 (BANK1)773/2,768 (27.9)987/3,612 (27.3)1.03 (0.9–1.2)
rs509749 (LY9)1,530/3,270 (46.8)1,777/3,810 (46.6)1.03 (0.9–1.2)
rs17435 (MECP2)   
 Women461/2,252 (20.5)445/2,224 (20.0)1.05 (0.9–1.2)
 Men71/372 (19.1)100/622 (16.1)1.23 (0.9–1.7)

Statistical analysis.

Tests for Hardy-Weinberg equilibrium in control samples were done with Haploview (21), using a threshold of 0.05, without correction for multiple testing. Other statistical analyses were done using a customized version of the Statistica 7.0 program (StatSoft, Tulsa, OK). Comparison of cases and controls was done with allele frequency data using chi-square tests on 2 × 2 contingency tables. These analyses were also conducted after stratifying the samples by sex and available clinical data. Results from the stratified analyses were subjected to Bonferroni correction. Univariate logistic regression models were used to test the fit to the data of additive, recessive, and dominant genetic models. Statistical power was estimated using “power and sample size calculations” software (22). Meta-analysis was done with our data and data from previous reports, using a fixed-effects model with R software (http://www.r-project.org/). Heterogeneity of the effect size between studies was assessed with the I2 statistic for inconsistency and Cochran's Q statistic. Samples from Hospital Universitario La Paz and Hospital Clinico San Carlos have already been used for the analysis of 2 genetic factors included in this project: BANK1 and STAT4. To avoid data duplication, we excluded these samples from the relevant analysis.

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

A total of 1,635 patients with RA and 1,906 control subjects of Spanish ancestry were available for study. The genotyping call rate across the 10 SNPs was 99.9%, and genotypes in the control subjects were in Hardy-Weinberg equilibrium.

We observed association of the STAT4 SNP rs7574865 with RA (odds ratio [OR] 1.14, 95% confidence interval CI 1.0–1.3, P = 0.028) but no significant differences between patients with RA and control subjects for the other 8 autosomal SNPs or for the MECP2 SNP, which is located in the X chromosome and was analyzed separately in female and male individuals (Table 2). All genotype comparisons were concordant with an additive genetic model and gave results similar to the allele comparisons (results not shown). The power of these analyses was enough to detect genetic factors showing an OR between 1.20 for the ITGAM SNP and 1.15 for the LY9 SNP (α = 0.05 and 1 – β = 0.80), which were the SNPs with minimum and maximum power, respectively.

Analysis was also done after stratification of samples according to sex and clinical data, which included anti–cyclic citrullinated peptide antibodies, rheumatoid factor, the shared epitope, rheumatoid nodules, radiographic changes, sicca symptoms, and interstitial pneumonitis. Most of these comparisons showed similar results across strata (detailed data are available from the corresponding author). The most remarkable difference was the decreased frequency of the minor allele of rs13277113 in the C8orf13–BLK locus among patients with RA and sicca symptoms (14.6% versus 24.2% in patients without these symptoms [P = 0.0005] and versus 22.4% in controls [P = 0.003]) (Table 3). These differences remained significant after correction for the number of tests performed (Table 3). However, this result should be taken with caution given the small number of patients in this group (n = 134) and association with a different allele than that in patients with SLE. There were other differences between strata (Table 3), but they were significant in the analyses only previous to Bonferroni correction.

Table 3. Significant differences in stratified analyses of patients with RA, according to sex and clinical features*
 Comparisons between patient strataComparisons with controls
SNP (locus)MAF, no. (%)OR (95% CI)PPcorrOR (95% CI)PPcorr
  • *

    The presence of erosions was defined radiographically. RA = rheumatoid arthritis; SNP = single-nucleotide polymorphism; MAF = minor allele frequency; OR = odds ratio; 95% CI = 95% confidence interval; NS = not significant; RF = rheumatoid factor.

  • Bonferroni correction.

rs6445975 (PXK)       
 Women541/2,252 (24.0)1.10 (0.9–1.3)NSNS1.19 (1.0–1.4)0.015NS
 Men168/752 (22.3)   0.87 (0.7–1.1)NSNS
rs13277113 (C8orf13–BLK)       
 Sicca syndrome39/268 (14.6)0.53 (0.4–0.8)0.00050.0080.59 (0.4–0.8)0.0030.04
 No sicca syndrome356/1472 (24.2)   1.11 (1.0–1.3)0.16NS
rs509749 (LY9)       
 Shared epitope330/650 (50.8)1.41 (1.1–1.8)0.0030.051.18 (1.0–1.4)0.05NS
 No shared epitope227/538 (42.2)   0.84 (0.7–1.0)0.05NS
rs10798269 (1q25.1)       
 RF positive673/2,184 (30.8)0.85 (0.7–1.0)0.06NS0.88 (0.8–1.0)0.02NS
 RF negative283/822 (34.4)   1.03 (0.9–1.2)NSNS
rs2304256 (TYK2)       
 Erosions565/2,108 (26.8)1.19 (1.0–1.4)0.06NS0.99 (0.9–1.1)NSNS
 No erosions209/888 (23.5)   0.83 (0.7–1.0)0.04NS

Finally, we combined our results with previous data including data available only as supplementary information and data imputed from the available genotypes but not directly tested in previous studies. These analyses included, at least, an additional data set for all of the SNPs except MECP2 SNP rs17435 and ITGAM SNP rs1143679, for which no previous information was found. The meta-analysis confirmed the results observed in our sample collection: the one SNP that is clearly associated with RA is the STAT4 SNP (Table 4), for which >35,000 subjects have already been studied. Results for this SNP showed significant heterogeneity of effects between studies (I2 = 53.0%, P for Cochran's Q statistic [PQ] = 0.007) that was not attributable to the inclusion of our results (I2 = 53.2%, PQ = 0.008). This heterogeneity prevents a sound estimate of the summary effect size, but it does not question association of the STAT4 SNP with RA, because all studies showed an OR of >1.0 for this SNP. Two other SNPs showed borderline association with RA: the SNP in the C8orf13–BLK locus for which ∼8,000 subjects have been analyzed (P = 0.03) and the SNP in PXK, with data from >11,000 subjects (P = 0.04).

Table 4. Combined analysis of allele frequency differences for SLE-associated SNPs obtained in the current and previous studies (including imputed genotypes) comparing patients with RA and controls*
SNP (locus)No. of RA patientsNo. of controlsOR (95% CI)PRef.
  • *

    SLE = systemic lupus erythematosus; SNPs = single-nucleotide polymorphisms; RA = rheumatoid arthritis; OR = odds ratio; 95% CI = 95% confidence interval; NS = not significant.

  • Analysis of heterogeneity between studies was possible only for STAT4, for which significant heterogeneity was observed (I2 = 53.0%, P for Cochran's Q statistic [PQ] = 0.007).

  • Data from refs.39 and46 overlap, and only the largest set from ref.39 was used here.

  • §

    I2 = 53.2%, PQ = 0.008, not including data from the present study.

rs7574865 (STAT4)17,56518,3151.25 (1.2–1.3)1.1 × 10−3431, 38–46
rs7574865 (STAT4)§16,18116,5091.26 (1.2–1.3)2.9 × 10−3431, 38–46
rs13277113 (C8orf13–BLK)3,4944,5621.09 (1.0–1.2)0.0331
rs2304256 (TYK2)3,1612,7630.94 (0.9–1.0)NS18
rs10798269 (1q25.1)5,0066,6940.97 (0.9–1.0)NS31, 46
rs4963128 (KIAA1542)5,0146,6941.01 (1.0–1.1)NS31, 46
rs6445975 (PXK)5,0176,6951.06 (1.0–1.1)0.0431, 46
rs17266594 (BANK1)3,5503,9600.97 (0.9–1.0)NS37
rs509749 (LY9)5,0146,6911.04 (1.0–1.1)NS31, 46

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

During the last decades, evidence supporting a common genetic component for a diversity of autoimmune diseases has been continuously investigated (10, 11, 15, 16, 23). The genetic factors that were associated with a certain disease were considered good candidates for the other diseases in the group, and they were systematically examined. This practice could have introduced a bias supporting the commonality hypothesis, because extensive investigation of the same factors in multiple diseases could have led to some associations just by chance. This is likely in the context of the lack of reproducibility that until recently characterized genetics studies of complex diseases (24). Accordingly, conflicting results have been the rule. Nowadays, genetics studies have achieved a high degree of reproducibility due to a significant increase in sample size and to more strict thresholds for claiming association (25). In addition, association across the genome has been analyzed in an unbiased manner (25). These developments gave us the opportunity to address the question of genetic commonality between RA and SLE with a new set of solidly established genetic factors.

The results provide evidence against widespread sharing of genetic factors between RA and SLE. Of the 9 newly examined SLE genetic factors, none was found to be associated with RA. This lack of association persisted after a meta-analysis including previous data, which makes it unlikely that any of the 9 representative SNPs have an effect in RA similar to that observed in SLE.

In retrospect, it could be contended that our findings were expected, given the identification of these genetic factors in the SLE genome-wide association study (26–28) and not in the RA genome-wide association study (29–31). However, the lack of significant association in a genome-wide association study does not mean the lack of any association, because genome-wide association studies apply strict thresholds for significance, and coverage of the different SLE and RA genome-wide association studies was not identical (i.e., only 1 of the 10 SNPs studied here was included in the largest RA genome-wide association study [31]). In addition, 3 SLE genetic factors explored here, TYK2 (32), MECP2 (33), and LY9 (34), were not found in the SLE genome-wide association study (although the first 2 factors have been independently confirmed [19, 35]). Therefore, our specific analysis together with the meta-analysis including previous data allow for a more conclusive assessment that reinforces the SLE specificity of these genetic factors in relation to RA.

Our study has enough power to detect effect sizes corresponding to an OR between 1.15 and 1.20 for the SNPs in autosomes and 1.24 for the MECP2 SNP in chromosome X. This is in the range of the weakest detectable effects by analysis of the largest RA sample collections, such as the North American Rheumatoid Arthritis Consortium family collection and the Epidemiological Investigation of Rheumatoid Arthritis (36). Combining our results with those from previous studies (including imputed data) increased the power. Two weak association signals with ORs of 1.09 and 1.06 were detected in the C8orf13–BLK and PXK loci, respectively. It is therefore possible that SNP effects below these sizes could have escaped our detection. In this case, the eventual RA-associated SNPs would be of lower significance for RA than for SLE disease, in which the reported ORs for these SNPs have been larger than 1.19, with the exception of single reports that did not replicate the LY9 and TYK2 associations (19, 35). It is also possible that different polymorphisms in the same gene are involved in different diseases, as indicated by a recent report of association of BANK1 SNPs with RA in a study that did not show association with the rs17266594 SNP that we examined here (37).

Another possibility we have addressed is that the sharing of genetic factors is restricted to specific subgroups of patients with RA, perhaps with more similarity with SLE. This is what happens with polymorphisms in IRF5 that are associated with SLE and with a not-fully-defined subgroup of patients with RA (17, 18). However, our analysis did not identify any clear association of this type. There were some association signals in the stratified analyses, but only differences in the C8orf13–BLK SNP in patients with sicca syndrome remained significant after correction. Unfortunately, no stratified data were available in previous reports that would allow us to draw more sound conclusions by increasing statistical power through meta-analysis.

In summary, the genetic factors in the HLA, PTPN22, 6q23, IRF5, and STAT4 loci are shared by SLE, RA, and other immune-mediated diseases (5, 6). However, none of the other 9 recently identified SLE genetic factors examined here showed association with RA. These results indicate that the genetic component of each of these 2 diseases contributes significantly to the different disease phenotypes. This is an idea that is not new but that lacked experimental support (10). The magnitude of the differential component cannot be estimated yet, because a large fraction of the heritability of RA and SLE has not been defined. More precise estimates will be possible once the causal variants of each of the genetic factors are identified. It is reasonable to expect that further genetic research will provide a clearer definition of what molecules and pathways are shared or specific in the pathogenesis of RA and SLE, allowing for better management of these 2 rheumatic diseases.

AUTHOR CONTRIBUTIONS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

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. Gonzalez 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. Suarez-Gestal, Gonzalez.

Acquisition of data. Dieguez-Gonzalez, Perez-Pampin, Pablos, Navarro, Narvaez, Marenco, Herrero-Beaumont, Fernandez-Gutierrez, Lamas, Rodriguez de la Serna, Ortiz, Carreño, Cañete, Caliz, Blanco, Balsa, Gomez-Reino, Gonzalez.

Analysis and interpretation of data. Suarez-Gestal, Calaza, Gomez-Reino, Gonzalez.

Acknowledgements

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES

We thank the sample donors for their generous collaboration.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES