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

Genetic factors account for an estimated 45–58% of the variance in joint destruction in rheumatoid arthritis (RA). The serine proteinase granzyme B induces target cell apoptosis, and several in vitro studies suggest that granzyme B is involved in apoptosis of chondrocytes. Serum levels of granzyme B are increased in RA and are also associated with radiographic erosions. The aim of this study was to investigate GZMB as a candidate gene accounting for the severity of joint destruction in RA.

Methods

A total of 1,418 patients with 4,885 radiograph sets of the hands and feet from 4 independent cohorts were studied. First, explorative analyses were performed in 600 RA patients in the Leiden Early Arthritis Clinic cohort. Fifteen single-nucleotide polymorphisms (SNPs) tagging GZMB were tested. Significantly associated SNPs were genotyped in data sets representing patients from the Groningen, Sheffield, and Lund cohorts. In each data set, the relative increase in the annual rate of progression in the presence of a genotype was assessed. Data were summarized in a meta-analysis. The association of GZMB with the RNA expression level of the GZMB genomic region was tested by mapping expression quantitative trait loci (QTLs) on 1,469 whole blood samples.

Results

SNP rs8192916 was significantly associated with the rate of joint destruction in the first cohort and in the meta-analysis of all data sets. Patients homozygous for the minor allele of rs8192916 had a higher rate of joint destruction per year compared with other patients (P = 7.8 × 10−4). Expression QTL of GZMB identified higher expression in the presence of the minor allele of rs8192916 (P = 2.27 × 10−5).

Conclusion

SNP rs8192916 located in GZMB is associated with the progression of joint destruction in RA as well as with RNA expression in whole blood.

Rheumatoid arthritis (RA) is an autoimmune disorder that affects 0.5–1% of the population and is associated with significant morbidity, disability, and cost to society. Radiographic joint destruction reflects the cumulative burden of inflammation and is considered to be an objective measure of the severity of RA (1). The degree of joint destruction varies significantly between patients. The processes behind this difference are incompletely understood. Inflammatory markers and autoantibodies are potent risk factors for joint destruction but explain ∼30% of the variance in joint destruction (2). Results of a twin study suggested that genetic factors influence the severity of joint destruction in RA, and a recent study in the Icelandic RA population estimated the heritability rate of joint destruction to be ∼45–58% (3, 4). Hence, to increase understanding of the processes mediating joint destruction, it would be beneficial to study genetic variants that could predispose to a severe outcome of RA.

Here, we describe a candidate gene study of the association of GZMB with the rate of joint destruction. Granzyme B is a serine protease found in the lytic granules of natural killer (NK) cells and cytotoxic T lymphocytes. When granzyme B is secreted into the interspace between the NK cell and the target cell, it triggers cell death by apoptosis (5, 6). In vitro studies have shown that granzyme B has enzymatic activity for the cleavage of aggrecan proteoglycans from cultured cartilage matrix and whole cartilage explants (7, 8); loss of cartilage proteoglycans is an early event in the course of destructive arthritis (9). In addition, studies of cartilaginous tissue demonstrated the presence of many granzyme B–positive cells among the chondrocytes in the pannus lesion (8). Furthermore, the levels of granzyme B were shown to be increased in the synovial fluid of patients with RA compared with that of healthy control subjects (10), and increased serum levels were associated with early development of radiographic erosions (11). Finally, a genetic variant (rs854350) in GZMB is reported to associate with susceptibility to RA (12).

These data led us to hypothesize that genetic variants in GZMB are associated with the severity of joint destruction in RA. We tested this hypothesis using 4 data sets comprising data for European RA patients for whom longitudinal radiography data on joint destruction were available. All data sets included patients in whom the diagnosis was made during a period when treatment strategies were less aggressive and less controlled than they are currently. These conservative treatment strategies made the data sets suitable for the present study, because the natural disease course was less inhibited. To further investigate the findings, the single-nucleotide polymorphisms (SNPs) associated with progression of joint destruction were also tested for their association with RNA expression.

PATIENTS AND METHODS

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

Study population.

Four data sets comprising data for adult European patients with RA were studied. RA was defined according to the 1987 American College of Rheumatology criteria (13) in all data sets except the Lund data set, for which the 1958 criteria for RA (14) were used. Radiographs of both the hands and feet were available for all patients (Table 1). All patients provided informed consent, and approval was obtained from the local ethics committee for each study.

Table 1. Characteristics of the patients represented by each data set*
CharacteristicLeiden EAC (n = 600)Groningen (n = 275)Sheffield (n = 396)Lund (n = 147)
  • *

    SHS = Sharp-van der Heijde score; anti-CCP = anti–cyclic citrullinated peptide.

  • Data for the Leiden Early Arthritis Clinic (EAC), Groningen, and Lund cohorts were obtained from baseline onward. Data for the Sheffield cohort were collected once during the disease period; the mean disease duration was 15 years (range 3–65 years).

Year of diagnosis, range1993–20061945–20011938–20031985–1989
Followup, years714Not applicable5
Total no. of radiograph sets2,846862396781
Scoring methodSHSSHSLarsenLarsen
Female sex, no. (%)412 (69)194 (71)290 (73)98 (67)
Age at diagnosis, mean ± SD years56 ± 1649 ± 1346 ± 1351 ± 12
Anti-CCP positive, no. (%)323 (55)160 (80)302 (79)114 (80)
Leiden Early Arthritis Clinic (EAC) cohort.

This data set represented 600 patients with early RA from the western part of The Netherlands, who were included in the Leiden EAC between 1993 and 2006 (2). Patients were included at the time of diagnosis and were followed up annually. Radiographs were obtained at baseline and at yearly followup visits for 7 years. A total of 2,846 sets of radiographs of the hands and feet were available. All radiographs were chronologically scored by one experienced reader who was unaware of the genetic or clinical data, using the Sharp/van der Heijde (SHS) scoring method for the hands and feet (15). A total of 499 randomly selected radiographs were scored twice. The within-reader intraclass correlation coefficient (ICC) was 0.91. Treatment of the patients could be divided into 3 time periods. Patients included in 1993–1995 were initially treated with nonsteroidal antiinflammatory drugs, patients included in 1996–1998 were initially treated with chloroquine or sulfasalazine, and patients included after 1999 were promptly treated with methotrexate or sulfasalazine.

Groningen cohort.

The second set of data represented 275 RA patients from the northern part of The Netherlands, in whom RA was diagnosed between 1945 and 2001. The duration of followup after diagnosis was limited to 14 years. The mean number of radiograph sets (hands and feet) per patient was 3.1 (maximum of 8 radiographs per patient). The total number of radiograph sets was 862. The radiographs were scored chronologically by 1 of 2 readers, using the SHS method. The ICC within readers was >0.90 and between readers was 0.96. In this data set, patients in whom RA was diagnosed before 1990 had, on average, a more progressive course of joint destruction compared with patients in whom RA was diagnosed after 1990. This difference in joint destruction progression is consistent with the approach of initiating disease-modifying antirheumatic drug (DMARD) treatment early, which was introduced in the 1990s.

Sheffield cohort.

The third data set represented 396 RA patients from the Sheffield, UK area. RA patients for whom radiographs were available were recruited from the Rheumatology Department of the Royal Hallamshire Hospital in Sheffield between 1999 and 2006 (16). RA patients were assessed once during their disease course. The mean ± SD disease duration at the time of the assessment was 15 ± 11 years (range 3–65 years). Radiographs of the hands and feet were scored by one reader using a modification of the Larsen score (17). Ten percent of the radiographs were scored twice to quantify intraobserver variation, using a weighted kappa value of 0.83 (16).

Lund cohort.

The fourth data set represented 183 Swedish patients with early RA who were prospectively followed up yearly over 5 years; radiographs and DNA were available for 147 of these patients (18, 19). Patients were recruited from primary care units in the Lund area from 1985 to 1989. Radiographs of the hands and feet were obtained at the start of the study and then annually for 5 years, resulting in a total of 781 sets of radiographs. Radiographs were scored chronologically by one of two readers, according to the Larsen method (20). The ICC between readers, as determined based on 105 radiographs, was 0.94. During the inclusion period, immediate DMARD therapy was not common, and only half of the patients were receiving any DMARD at the 5-year followup. The DMARDs used most commonly were chloroquine, D-penicillamine, sodium aurothiomalate, and auranofin (18).

SNP selection and genotyping.

The region of GZMB plus the haplotype block upstream and downstream of the gene were tagged using the pairwise Tagger algorithm developed by de Bakker et al (21) and implemented in Haploview (22). Two SNPs (rs8192917 and rs2236338) were known to be amino acid–changing SNPs. One intergenic SNP, rs854350, had a significant association with RA susceptibility in the data set of the Wellcome Trust Case Control Consortium (12). These 3 SNPs were force included. Pairwise-tagging SNPs were selected from the CEPH/CEU HapMap data set (phase II, release 21, NCBI build 35) using Haploview software (minor allele frequency [MAF] >0.05, pairwise r2 > 0.8). Sixteen SNPs captured GZMB. Multiplex SNP arrays were designed using an Illumina GoldenGate platform, according to the protocols recommended by the manufacturer. One SNP could not be designed (rs1951601), but a good proxy (rs12433772; r2 = 0.90) was typed instead. The final SNP selection and the linkage disequilibrium (LD) information are shown in Figure 1.

thumbnail image

Figure 1. Linkage disequilibrium structure between 16 tag single-nucleotide polymorphisms (SNPs) in GZMB. The data are for 600 patients with rheumatoid arthritis (RA) from the Leiden Early Arthritis Clinic cohort. The numbers present the r2 value between the SNPs. The variations in shading refer to D′. SNPs rs2236338 and rs8192917 are amino acid–changing SNPs, and rs854350 was significantly associated with RA susceptibility in the Wellcome Trust Case Control Consortium data set (solid arrows). These SNPs were force-included. One SNP, rs1951601, could not be designed, but a good proxy existed (rs1243372; r2 = 0.90). One SNP, rs8192920, was not analyzed because of a low success rate for typing. Open arrows indicate SNPs that were significantly associated with joint destruction.

Download figure to PowerPoint

Software supplied by Illumina was used to automatically identify the genotypes. Each 96-well plate consisted of one positive and one negative control. Clusters were evaluated, and all doubtful calls were checked. After manually evaluating the spectra of each cluster, the genotypes were accepted, recalled, or rejected. At least 12% of the genotypes were assessed in duplicate, with an error rate of <1% for all SNPs. One SNP, rs8192920, had a success rate of 75% and was therefore excluded from further analysis. The remaining SNPs were typed with a success rate of >95% (additional information is available from the corresponding author). None of the SNPs deviated from Hardy-Weinberg equilibrium (P < 0.001).

The SNP that was significantly associated with joint destruction in the first cohort was genotyped in the other 3 cohorts. The SNP was genotyped as a part of multiplex SNP arrays designed with Sequenom iPLEX, according to the protocols recommended by the manufacturer. Software supplied by the same manufacturer was used to automatically identify the genotypes. Each iPLEX consisted of at least 9 positive controls and 9 negative controls. All doubtful calls were checked manually; DNA samples in which >30% of the SNPs failed were excluded from analysis (n = 31). At least 5% of the genotypes were assessed in duplicate, with an error rate of <1%. The success rates were all >95%. None of the SNPs deviated from Hardy-Weinberg equilibrium.

Measuring and analyzing GZMB expression levels.

The SNP that remained significant after meta-analysis was tested for its association with RNA expression of the genomic region of GZMB, by mapping expression quantitative trait locus (eQTL) on a data set representing peripheral blood samples obtained from 1,469 unrelated individuals (23, 24). RNA expression was tested using probes with a midpoint position <250 kb from the tested SNP, thus testing for a cis effect. Expression QTL effects were determined using Spearman's rank correlation coefficient. If an association was observed, it was determined whether other SNPs in the region (n = 583) had a stronger cis-eQTL effect. If this was the case, the first analysis was repeated using linear regression with adjustment for the strongest associating SNP, in an effort to determine the eQTL effect of the assessed SNPs independently of LD with the most strongly associated SNP. Genotypes were imputed with HapMap2 (release 24), using Impute version 2. Only SNPs with a MAF >5% and a Hardy-Weinberg P value greater than 0.0001 were included for analysis. Correction for multiple testing was performed by controlling the false discovery rate at 5%, permuting the gene expression labels 100 times as previously described (25). Finally, to prevent false-positive findings due to primer polymorphisms, SNP–probe combinations were excluded from the cis-eQTL analysis when the 50-bp long expression probe mapped to a genomic location that contained a known SNP that was showing at least some LD (r2 > 0.1) with the cis SNP.

Statistical analysis.

Associations between genotypes and radiographic joint destruction were analyzed in 2 phases. First, an explorative analysis was performed in the Leiden EAC data set. In this data set, the tagged SNPs were tested both additively and recessively. Because phase 1 was an explorative phase, no correction for multiple testing was applied, and SNPs with a P value less than or equal to 0.05 were studied in phase 2.

For analyses involving the Leiden EAC, Groningen, and Lund data sets, a multivariate normal regression model for longitudinal data was used, with the radiographic score as the response variable. This method is used to analyze all repeated measurements at once and takes advantage of the correlation between these measurements. This model is similar to a linear mixed-effects model, but no random-effects model is added (26). To model the correlation over time, a heterogeneous first-order autoregressive matrix was used, which assumes a stronger correlation for measurements obtained over a shorter period of time compared with those obtained over a longer period of time. The effect of time was entered as a factor in the model, to capture properly the mean response profile over time.

To test for an association with the rate of joint destruction, an analysis with the SNP and its interaction with time (as a linear variable) in the model was conducted. Adjustment variables were entered based on their univariate association with joint destruction. In the Leiden EAC data set, adjustments were made for age, sex, and the described treatment periods. The Groningen data set was adjusted for age and the period before or during 1990 and the period after 1990, as a proxy for DMARD therapy. The analysis of the Lund data set was adjusted for age only, because sex and treatment were not associated with joint destruction in this data set.

In the Sheffield data set, each patient had a set of radiographs of the hands and feet at one time point. To make the scores comparable with those of the other data sets, the estimated yearly progression rate was calculated (27). This was achieved by dividing the total Larsen score by the number of disease-years at the time of radiography, resulting in an estimation of the rate of joint destruction. Information regarding disease duration at the time of radiography was available for 391 patients. The SNP association was tested in a linear regression analysis, with the log-transformed estimated yearly progression rate as the outcome variable. No adjustments were applied, because none of the tested variables was univariately significantly associated with joint destruction. The analyses were performed using SPSS version 17.0.

In all data sets, the radiographic scores were log-transformed to obtain a normal distribution. Because the analyses were performed on the log scale, the resulting coefficient on the original scale indicates the fold increase in joint destruction per year of followup: over a followup period of n years, the coefficient increases to the power of n.

In the present study, the power to detect genetic effects is a function of the number of patients and the number of measurements per patient studied. As shown previously, the precision of the estimate increases steadily with increasing numbers of radiographs per patient (28). All 3 data sets studied to verify the results obtained in phase 1 contained (individually and combined) fewer radiographs than the initial data set. Consequently, the power to replicate findings in each data set individually as well as in the 3 replication data sets together could be limited due to the large number of radiographs in the discovery data set. Because of differences in study designs, the separate data sets could not be combined in one analysis directly. Therefore, the SNPs in each data set were tested separately, thereby taking advantage of the specific data set characteristics, and a meta-analysis of the results was performed subsequently to determine the association of the SNPs with the rate of joint destruction. Because the parameters in all data sets reflect the relative increase in the rate of joint destruction per year, the estimates of the individual data sets were pooled in a meta-analysis (29). A fixed-effects meta-analysis (30) with inverse-variance weighting was performed using Stata version 10.1.

RESULTS

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

Phase 1 (SNP identification).

GZMB was tagged by 16 SNPs, one of which was not analyzed because of a low typing success rate. From the 15 analyzed SNPs, 2 (rs8192916 and rs12433772) were significantly associated with joint destruction in the Leiden EAC data set (Table 2). Patients homozygous for the minor allele of rs8192916 had an average 1.05-fold (range 1.02–1.08-fold) increase in the rate of joint destruction per year compared with the rate in the other patients (P = 1.2 × 10−3) (Figure 2). Because the estimated rate of progression increases by the power of the number of years, a coefficient of 1.05 per year resulted in a 1.41-fold increased rate of joint destruction over 7 years. SNP rs12433772 was highly linked to rs8192916, and because the effect of rs8192916 was more evident at the subsequent time points, this SNP was prioritized for the phase 2 analysis.

Table 2. GZMB SNPs associated with progression of joint destruction in the Leiden data set*
SNPCoordinateTested modelMAFCoefficient95% CIP
  • *

    For the tagging process, NCBI build 35, dbSNP b125 was applied. To make all coordinates in the study consistent (single-nucleotide polymorphisms [SNPs] and probes), NCBI build 36 was used to report the locations. The coefficient of the additive (ADD) test represents the relative increase in joint destruction per year for 1 minor allele compared with no minor alleles. The coefficient of the recessive (REC) test represents the relative increase in joint destruction per year for 2 minor alleles compared with 1 or no minor alleles, on the normal (non-log) scale. A coefficient of 1.05 per year indicates a 5% higher rate of joint destruction; this implies a 41% higher rate of joint destruction over a 7-year period. MAF = minor allele frequency; 95% CI = 95% confidence interval.

rs819291624174200ADD0.421.011.00–1.030.11
  REC 1.051.02–1.081.2 × 10−3
rs1243377224190515ADD0.381.021.00–1.040.02
  REC 1.061.03–1.092.9 × 10−4
thumbnail image

Figure 2. Median Sharp/van der Heijde (SHS) scores during followup of rheumatoid arthritis patients with different rs8192916 genotypes in the Leiden Early Arthritis Clinic cohort. The rate of joint destruction per year was 1.05-fold greater in patients homozygous for the minor allele of rs8192916 (1.41-fold increased rate over 7 years).

Download figure to PowerPoint

Phase 2 (meta-analysis).

SNP rs8192916 was tested in a recessive model in all 3 replication data sets. As expected, due to insufficient power of the replication cohorts, the 95% confidence intervals (95% CIs) in each of the 3 cohorts separately all included 1 (Figure 3). An inverse-variance weighting meta-analysis was used to analyze rs8192916 in all 1,418 patients. SNP rs8192916 remained significant (P = 7.8 × 10−4), thereby estimating a higher rate of joint destruction per year in patients homozygous for the minor allele compared with the other patients (Figure 3). This association remained significant (P = 1.2 × 10−2) after conservative correction (Bonferroni adjustment) for testing 16 SNPs (the number of SNPs tested in phase 1).

thumbnail image

Figure 3. Results of the analyses of rs8192916 in all 4 cohorts and of the final meta-analysis. The effect sizes are the estimated relative progression rates per year for the presence of 2 minor alleles compared with 1 or no minor allele. A fixed-effects meta-analysis was applied to determine statistically significant effects; the generalizability of the effect is of less importance. As a result of this choice, the effect size of the meta-analysis should be considered carefully. Consequently, this method is less suitable to estimate the effect size overall. The estimated effect of the meta-analysis is shown in gray. Leiden-EAC = Leiden Early Arthritis Clinic.

Download figure to PowerPoint

When the association of rs8192916 with the rate of joint destruction was stratified for anti–citrullinated protein antibody (ACPA) status, the association was present in both subgroups (effect size 1.05 [95% CI 1.02–1.08], P = 1.98 × 10−3 and effect size 1.03 [95% CI 0.99–1.08], P = 5.40 × 10−2 for ACPA-negative and ACPA-positive patients, respectively).

Measuring and analyzing GZMB expression levels.

To further substantiate the observed association of rs8192916 with joint destruction, the effect of rs8192916 on RNA expression was studied by cis-eQTL mapping in peripheral blood samples obtained from 1,469 unrelated individuals (22).

SNP rs8192916 was significantly correlated with RNA expression of 4 different genes: GZMB, CTSG, CBLN3, and an unnamed gene with accession no. AK056368 (P = 1.38 × 10−5 to 6.14 × 10−6) (Table 3). For all 4 genes, a SNP other than rs8192916 was more strongly correlated with RNA expression. To assess whether the SNP associated with joint destruction, rs8192916, had an effect on gene expression independently of the effect of the most strongly associated SNPs, the analyses were subsequently adjusted for the strongest SNP. After this correction was performed, rs8192916 was still associated with RNA expression of the GZMB probe (P = 0.015) (Table 3).

Table 3. Significant correlations of rs8192916 with RNA cis-eQTLs in peripheral blood*
Expressed geneProbe midpointP, corrected for strongest SNPFalse discovery rateStrongest associating SNPPCorrected P
  • *

    Information regarding the location of probes is based on NCBI 36.3 build.

  • Single-nucleotide polymorphisms (SNPs) in the region most strongly correlated with RNA expression.

  • The association of rs8192916 with cis–expression quantitative trait loci (eQTL) was corrected for the strongest correlated SNP. SNP rs8192916 was correlated with RNA expression of several probes. Four probes, covering 4 different genes, were significantly correlated with rs8192916. These 4 correlations were further corrected for the SNPs that most strongly correlated with the RNA expression.

GZMB241716652.27 × 10−53.03 × 10−5rs22363373.98 × 10−1190.015
CTSG241139351.38 × 10−51.38 × 10−5rs128785788.91 × 10−630.115
CBLN3239670772.27 × 10−54.54 × 10−5rs22736291.79 × 10−690.057
AK056368240160936.14 × 10−62.46 × 10−6rs128960863.46 × 10−140.752

DISCUSSION

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

The variance in joint destruction between patients with RA is considerable, and the mechanisms driving these differences are thus far scarcely understood. We performed a candidate gene study to investigate the association of genetic variants in GZMB with joint destruction. GZMB (14q11.2) was chosen as a candidate gene, because granzyme B is involved in inflammation, and data are emerging that granzyme B could be relevant to joint destruction in RA (5–8, 10, 11). We tested the association of SNPs tagging GZMB with the rate of joint destruction in one data set and subsequently tested the significant SNP in 3 other data sets. Next, a meta-analysis combining the radiographic data of all 1,418 patients was performed. One SNP, rs8192916 (situated 9 kb upstream of GZMB), was observed to associate significantly with progression of joint destruction. The minor allele of rs8192916 was associated with a higher rate of joint destruction. We further observed that carrying this minor allele of rs8192916 was correlated with higher RNA expression of GZMB in whole blood.

The present study uniquely combines 4 data sets representing patients with a similar European ethnicity who were treated in the prebiologic agent era. Hence, the radiographic progression rate of the patients studied here are more reflective of the natural course of RA compared with that of recently treated patients. In some data sets, patients were included over a wide time span; as treatment strategies changed over time, these patients received different treatments. Because different treatment regimens are associated with our outcome of interest (joint destruction), the analyses were adjusted for treatment, when relevant. Further studies on different patient populations as well as those treated with biologic agents would be informative.

Replication data sets are ideally larger than the initial data set; because the effect sizes in the discovery data set are likely to be upwardly biased, a smaller effect size is expected at a replication stage. Unfortunately, relatively few large prospective data sets exist in which both radiographs and DNA are available for conventionally treated patients. The number of patients and the number of radiographs in each data set separately were insufficient to allow well-powered analyses. Also, the replication data sets combined contained fewer radiographic measurements than were included in phase 1. Consequently, the data available for the phase 2 analysis were expected to be underpowered to replicate findings individually. Therefore, data were summarized in an inverse-variance weighting meta-analysis, which showed a significant result for rs8192916.

To further substantiate the findings for GZMB, RNA expression was studied and revealed a strong signal for rs8192916. To confirm that this correlation was not driven by a linked SNP, we sought to identify other SNPs in the region that also associate with RNA expression. SNP rs2236337 had the strongest correlation with GZMB expression. However, LD between rs2236337 and rs8192916 was low (r2 = 0.08). Moreover, the fact that after correction for this SNP, rs8192916 remained significant suggests that the correlation of rs8192916 with GZMB expression is likely independent of the influence of rs2236337.

In the current study, GZMB was tagged by pairwise SNP selection, with an MAF of >0.05 and a pairwise r2 value of >0.8. Inherent to this method, rare variants are not selected. Therefore, based on the results of the current study, we cannot exclude the possibility that other variants in GZMB also associate with joint destruction.

Granzyme B, which is released by NK cells and T lymphocytes, can induce cell death in harmful cells, such as those that are virally infected or malignant (31). Recent findings have suggested a role for granzyme B in RA joint destruction as well; the number of granzyme B–positive cells is increased in synovium as well as among chondrocytes at the site of pannus lesions (8, 32).

The findings of the current study are consistent with those of previous studies. Possibly, the minor variant of rs8291916, or a genetic variant linked to this SNP, induces higher expression of granzyme B, leading to more apoptosis in chondrocytes and therefore more joint destruction. This would explain the previously observed association of higher serum granzyme B levels in patients with erosive disease (11). Another explanation could be the role of granzyme B in propagation of the inflammatory response; additional studies are needed to determine this.

In conclusion, using a candidate gene approach evaluating patients in 4 different cohorts, we observed associations between rs8192916 and an increased rate of joint destruction in RA and between rs8192916 and expression profiles of GZMB. Although evaluation of rs8192916 at the protein level was not performed, these data suggest that carriage of the risk allele may affect the function of GZMB.

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. Knevel 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. Knevel, Krabben, Wilson, van der Linden, Huizinga, van der Helm-van Mil.

Acquisition of data. Knevel, Wilson, Brouwer, Leijsma, Lindqvist, Daha, van der Linden, Saxne, van der Helm-van Mil.

Analysis and interpretation of data. Knevel, de Rooy, van der Linden, Tsonaka, Zhernakova, Westra, Franke, Houwing-Duistermaat, Toes, Huizinga, Saxne, van der Helm-van Mil.

Acknowledgements

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

We thank all the individuals who participated in this study.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. AUTHOR CONTRIBUTIONS
  7. Acknowledgements
  8. REFERENCES
  • 1
    Van der Heijde DM, van Riel PL, van Leeuwen MA, van 't Hof MA, van Rijswijk MH, van de Putte LB. Prognostic factors for radiographic damage and physical disability in early rheumatoid arthritis: a prospective follow-up study of 147 patients. Br J Rheumatol 1992; 31: 51925.
  • 2
    De Rooy DP, van der Linden MP, Knevel R, Huizinga TW, van der Helm-van Mil AH Predicting arthritis outcomes: what can be learned from the Leiden Early Arthritis Clinic? Rheumatology (Oxford) 2011; 50: 93100.
  • 3
    Van der Helm-van Mil AH, Kern M, Gregersen PK, Huizinga TW. Variation in radiologic joint destruction in rheumatoid arthritis differs between monozygotic and dizygotic twins and pairs of unrelated patients. Arthritis Rheum 2006; 54: 202830.
  • 4
    Knevel R, Grondal G, Huizinga TW, Visser AW, Jonsson H, Vikingsson A. Genetic predisposition of the severity of joint destruction in rheumatoid arthritis: a population-based study. Ann Rheum Dis 2012; 71: 7079.
  • 5
    Trapani JA, Davis J, Sutton VR, Smyth MJ. Proapoptotic functions of cytotoxic lymphocyte granule constituents in vitro and in vivo. Curr Opin Immunol 2000; 12: 3239.
  • 6
    Lord SJ, Rajotte RV, Korbutt GS, Bleackley RC. Granzyme B: a natural born killer. Immunol Rev 2003; 193: 318.
  • 7
    Froelich CJ, Zhang X, Turbov J, Hudig D, Winkler U, Hanna WL. Human granzyme B degrades aggrecan proteoglycan in matrix synthesized by chondrocytes. J Immunol 1993; 151: 716171.
  • 8
    Ronday HK, van der Laan WH, Tak PP, de Roos JA, Bank RA, TeKoppele JM, et al. Human granzyme B mediates cartilage proteoglycan degradation and is expressed at the invasive front of the synovium in rheumatoid arthritis. Rheumatology (Oxford) 2001; 40: 5561.
  • 9
    Korb-Pap A, Stratis A, Muhlenberg K, Niederreiter B, Hayer S, Echtermeyer F, et al. Early structural changes in cartilage and bone are required for the attachment and invasion of inflamed synovial tissue during destructive inflammatory arthritis. Ann Rheum Dis 2012; 71: 100411.
  • 10
    Tak PP, Spaeny-Dekking L, Kraan MC, Breedveld FC, Froelich CJ, Hack CE. The levels of soluble granzyme A and B are elevated in plasma and synovial fluid of patients with rheumatoid arthritis (RA). Clin Exp Immunol 1999; 116: 36670.
  • 11
    Goldbach-Mansky R, Suson S, Wesley R, Hack CE, El-Gabalawy HS, Tak PP. Raised granzyme B levels are associated with erosions in patients with early rheumatoid factor positive rheumatoid arthritis. Ann Rheum Dis 2005; 64: 71521.
  • 12
    Wellcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature 2007; 447: 66178.
  • 13
    Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988; 31: 31524.
  • 14
    Ropes MW, Bennett GA, Cobb S, Jacox R, Jessar RA. 1958 revision of diagnostic criteria for rheumatoid arthritis. Bull Rheum Dis 1958; 9: 1756.
  • 15
    Van der Heijde D. How to read radiographs according to the Sharp/van der Heijde method. J Rheumatol 2000; 27: 2613.
  • 16
    Mewar D, Coote A, Moore DJ, Marinou I, Keyworth J, Dickson MC. Independent associations of anti-cyclic citrullinated peptide antibodies and rheumatoid factor with radiographic severity of rheumatoid arthritis. Arthritis Res Ther 2006; 8: R128.
  • 17
    Rau R, Herborn G. A modified version of Larsen's scoring method to assess radiologic changes in rheumatoid arthritis. J Rheumatol 1995; 22: 197682.
  • 18
    Fex E, Jonsson K, Johnson U, Eberhardt K. Development of radiographic damage during the first 5-6 yr of rheumatoid arthritis: a prospective follow-up study of a Swedish cohort. Br J Rheumatol 1996; 35: 110615.
  • 19
    Lindqvist E, Jonsson K, Saxne T, Eberhardt K. Course of radiographic damage over 10 years in a cohort with early rheumatoid arthritis. Ann Rheum Dis 2003; 62: 6116.
  • 20
    Larsen A, Dale K. Standardised radiological evaluation of rheumatoid arthritis in therapeutic trials. In: Dumonde DC, Jasani MK, eds. The recognition of anti-rheumatic drugs. Lancaster (UK): MTP Press; 1978. p. 28592.
  • 21
    De Bakker PI, Yelensky R, Pe'er I, Gabriel SB, Daly MJ, Altshuler D. Efficiency and power in genetic association studies. Nat Genet 2005; 37: 121723.
  • 22
    Barrett JC, Fry B, Maller J, Daly MJ. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005; 21: 2635.
  • 23
    Dubois PC, Trynka G, Franke L, Hunt KA, Romanos J, Curtotti A, et al. Multiple common variants for celiac disease influencing immune gene expression. Nat Genet 2010; 42: 295302.
  • 24
    Fehrmann RS, Jansen RC, Veldink JH, Westra HJ, Arends D, Bonder MJ, et al. Trans-eQTLs reveal that independent genetic variants associated with a complex phenotype converge on intermediate genes, with a major role for the HLA. PLoS Genet 2011; 7: e1002197.
  • 25
    Breitling R, Li Y, Tesson BM, Fu J, Wu C, Wiltshire T, et al. Genetical genomics: spotlight on QTL hotspots. PLoS Genet 2008; 4: e1000232.
  • 26
    Fitzmaurice GM, Laird NM, Ware JH. Applied longitudinal analysis. New York: John Wiley & Sons; 2004.
  • 27
    Strand V, Landewe R, van der Heijde D. Using estimated yearly progression rates to compare radiographic data across recent randomised controlled trials in rheumatoid arthritis. Ann Rheum Dis 2002; 61 Suppl 2: ii646.
  • 28
    Van der Helm-van Mil AH, Knevel R, van der Heijde D, Huizinga TW. How to avoid phenotypic misclassification in using joint destruction as an outcome measure for rheumatoid arthritis? Rheumatology (Oxford) 2010; 49: 142935.
  • 29
    Lassere M. Pooled metaanalysis of radiographic progression: comparison of Sharp and Larsen methods. J Rheumatol 2000; 27: 26975.
  • 30
    Lebrec JJ, Stijnen T, van Houwelingen HC. Dealing with heterogeneity between cohorts in genomewide SNP association studies. Stat Appl Genet Mol Biol 2010; 9: Article 8.
  • 31
    Trapani JA, Sutton VR. Granzyme B: pro-apoptotic, antiviral and antitumor functions. Curr Opin Immunol 2003; 15: 53343.
  • 32
    Saito S, Murakoshi K, Kotake S, Kamatani N, Tomatsu T. Granzyme B induces apoptosis of chondrocytes with natural killer cell-like cytotoxicity in rheumatoid arthritis. J Rheumatol 2008; 35: 193243.