ARTHRITIS & RHEUMATISM
Vol. 52, No. 1, January 2005, pp 354-360
© 2005, American College of Rheumatology
Association between stromal cell–derived factor 1
chemokine gene variant and radiographic progression
of rheumatoid arthritis
Stromal cell–derived factor 1 (SDF-1 or CXCLI2) is a
potent CXC chemokine involved in physiologic and pathologic
homing of a large variety of cells expressing its specific
receptor CXCR4 (1). Increased expression of SDF-1 has been
demonstrated in rheumatoid arthritis (RA) synovium, where it
is expressed by synoviocytes and presented on endothelium
(2–4). In addition to its potential role in cell recruitment,
SDF-1 produced by hyperplastic synoviocytes seems to be
involved in several processes relevant to the mechanisms of
bone and cartilage destruction in RA, including neoangiogen-
esis (4), synthesis of matrix metalloproteinases (5,6), and
osteoclastogenesis (7,8). CXCR4 inhibitors consistently de-
crease the severity of arthritis in a murine model (9). Although
SDF-1 inhibitors have not been tested in RA patients, we have
observed an indirect reduction in SDF-1 expression in the
synovial tissue of patients achieving a good clinical response to
anti–tumor necrosis factor therapy (10).
An allelic variant in the 3?-untranslated region (3?-
UTR) of the SDF-1 gene (801 G/A) has been described (11).
Although the functional role of these variants has not been
clearly established, several studies suggest that individuals with
different genotypes display functional changes in SDF-1 ex-
pression and function. SDF-1 interferes with cellular entry by
human immunodeficiency virus (HIV), and HIV-infected in-
dividuals carrying different SDF-1 3?-UTR alleles show differ-
ent rates of progression to acquired immunodeficiency syn-
drome (AIDS) or AIDS-related death (11). An association
between the SDF-1 3?-UTR AA genotype and earlier onset of
autoimmune diabetes has also been demonstrated (12). We
have studied whether SDF-1 3?-UTR alleles are associated
with RA susceptibility or severity by comparing a cohort of RA
patients with healthy controls and by analyzing potential
correlations between long-term radiographic progression and
SDF-1 3?-UTR alleles.
We selected from a large cohort of RA patients those
who were regularly followed up at our hospital rheumatology
unit for more than 10 years from RA onset. All patients met
the American College of Rheumatology (formerly, the Amer-
ican Rheumatism Association) 1987 revised criteria for RA
(13). We selected patients with available radiographs of the
hands and feet obtained a mean ? SD of 10 ? 2 years after RA
onset. DNA samples were obtained from 138 Spanish RA
patients and from 158 Spanish unrelated individuals randomly
selected among voluntary blood donors from the same geo-
Radiographs were analyzed in a blinded manner and
scored according to the modified Sharp/Van der Heijde
method (14) by a single observer. The joint erosion and joint
narrowing scores were added to generate the total radio-
graphic damage score. Intraobserver variability was evaluated
in a randomly selected sample of 26 radiographs.
Genomic DNA was isolated from peripheral blood
mononuclear cells and analyzed by polymerase chain reaction–
restriction fragment length polymorphism (PCR-RFLP) ana-
lysis as previously described (15). In selected samples, sequenc-
ing of the PCR product was performed to confirm the
identification of alleles determined by RFLP analysis.
Allele and genotype frequencies in RA patients and
healthy controls were compared by chi-square test. Analysis-
of-variance (ANOVA) was used to compare the radiographic
scores in the different groups. Patients’ characteristics were
compared by ANOVA or chi-square test as appropriate.
Intraobserver variability of radiographic scoring was analyzed
by the Pearson correlation coefficient.
The allele and genotype frequencies in patients and
controls are illustrated in Table 1. The distribution was con-
sistent with Hardy-Weinberg equilibrium. No significant dif-
ferences in allele and genotype frequencies of the SDF-1
3?-UTR A variant were detected between RA patients and
healthy unrelated controls.
Demographic variables, the proportions of rheumatoid
factor–positive patients, and the use of disease-modifying
antirheumatic drug therapy did not differ significantly between
the different genotype groups (Table 2). At 10 ? 2 years after
RA onset, patients with the SDF-1 3?-UTR AA homozygous
genotype had significantly higher erosion, narrowing, and total
modified Sharp/Van der Heijde scores than those of patients
with the GG or GA genotypes considered together (Table 3).
No significant differences in radiographic scores were found
between patients with the GG and GA genotypes, nor between
groups of patients carrying the A or G allele, indicating a
recessive model for the SDF-1 3?-UTR A variant. Intraob-
server variability in radiographic scoring, as analyzed by the
Pearson correlation coefficient, was extremely low (r ? 0.99
[95% confidence interval 0.94–1.4], P ? 0.001).
Although no significant differences were detected be-
tween the different genotype groups in the mean duration
from disease onset to radiographic analysis, we also compared
the annual rates of radiographic progression from RA onset to
the time of the last radiographic study in the different geno-
type groups. This analysis confirmed significantly higher an-
nual rates of radiographic progression of erosions, narrowing,
or total damage in patients with the SDF-1 3?-UTR AA
homozygous genotype compared with patients with the other
genotypes (Table 3).
During chronic inflammation, proinflammatory che-
mokines are locally induced by a variety of proinflammatory
stimuli and participate in the recruitment and activation of
leukocytes. SDF-1 is not a typical proinflammatory chemo-
polymorphism in RA patients and healthy unrelated controls*
Genotype and allele frequencies of the SDF-1 3?-UTR 801
(n ? 158)
(n ? 138)
* Values are the number (%) of patients with a given genotype or
allele. SDF-1 ? stromal cell–derived factor 1; 3?-UTR ? 3?-
untranslated region; RA ? rheumatoid arthritis.
kine; it is instead considered a homing chemokine involved in
cell traffic to bone marrow and lymphoid organs (1). However,
increased expression and endothelial display of SDF-1 in RA
synovium suggests that it participates in the pathogenesis of
the disease. In addition to its potential role in inflammatory
cell recruitment, several biologic actions of this chemokine
suggest that it can operate as a progression factor by partici-
pating in the mechanisms of pannus formation and bone and
cartilage destruction. SDF-1 is a proangiogenic factor in
multiple settings, including RA, in which it is displayed by
angiogenic vessels and contributes to the ex vivo angiogenic
activity of RA synovial fluid (4). SDF-1 may also contribute to
bone and cartilage degradation by inducing the release of
matrix metalloproteinases (5,6), and it may also contribute to
osteoclast development and activity (7,8).
Our data consistently suggest a role for SDF-1 gene
variation in RA progression rather than in RA susceptibility.
This suggests that early immune-mediated events are not
SDF-1 dependent, providing further support to the hypothesis
of potential participation of SDF-1 in the progression of
synovitis and tissue destruction. The relevance of SDF-1 in
disease progression has also been suggested by studies in
murine arthritis, in which therapy with a CXCR4 antagonist
improved arthritis progression after disease onset (9).
Although SDF-1 levels are increased in RA synovial
tissue (2,4,8), the mechanisms involved remain unclear. The
role of the 3?-UTR in regulating SDF-1 transcription or
messenger RNA stability has not been explored in synovio-
cytes. Our data, together with the results of studies in HIV and
diabetes, suggest that the SDF-1 3?-UTR AA genotype could
be linked to higher SDF-1 expression. Preliminary data re-
ported from a study of human fibroblasts transfected with a
reporter gene coupled in the 3?-UTR to the different SDF-1
3?-UTR variants show that the 801A allele is associated with
higher gene expression compared with the 801G allele (16).
In conclusion, our genetic data lend support to the
proposed hypothesis of the participation of SDF-1 in the
pathogenesis of RA and, more specifically, in the progression
of joint destruction. Further studies of the potential contribu-
tion of the SDF-1 3?-UTR to the regulation of SDF-1 expres-
sion in synovial fibroblasts may help to identify the mecha-
nisms that lead to SDF-1 overexpression in inflammatory
Supported by grants from Fundacio ´n de Investigacio ´n Me ´dica
Mutua Madrilen ˜a, Fundacio ´n Prevencio ´n e Investigacio ´n del SIDA
(FIPSE 3120/00, 36453/03), and Red de Investigacio ´n en SIDA. Dr.
Santiago’s work was supported by a grant from Abbott Laboratories to
Fundacio ´n Espan ˜ola de Reumatologı ´a. Ms Gonza ´lez’s work was sup-
ported by the FPI program from Ministerio de Educacio ´n y Ciencia.
Beatriz Joven, MD
Hospital 12 de Octubre
Nuria Gonza ´lez, MSc
Francisco Aguilar, MSc
Instituto de Salud Carlos III
Begon ˜a Santiago, PhD
Marı ´a Galindo, MD
Hospital 12 de Octubre
Jose ´ Alcamı ´, MD
Instituto de Salud Carlos III
Jose ´ L. Pablos, MD
Hospital 12 de Octubre
1. Murdoch C. CXCR4: chemokine receptor extraordinaire. Immu-
nol Rev 2000;177:175–84.
2. Nanki T, Hayashida K, El-Gabalawy HS, Suson S, Shi K, Girschick
HJ, et al. Stromal cell-derived factor-1-CXC chemokine receptor 4
interactions play a central role in CD4? T cell accumulation in
rheumatoid arthritis synovium. J Immunol 2000;165:6590–8.
3. Blades MC, Ingegnoli F, Wheller SK, Manzo A, Wahid S, Panayi
GS, et al. Stromal cell–derived factor 1 (CXCL12) induces mono-
cyte migration into human synovium transplanted onto SCID
mice. Arthritis Rheum 2002;46:824–36.
4. Pablos JL, Santiago B, Galindo M, Torres C, Brehmer MT, Blanco
801 genotype groups*
Characteristics of patients in the different SDF-1 3?-UTR
P† GG GAAA
Women, no. (%)
Age, mean ? SD years
RF positive, no. (%)
mean ? SD
Methotrexate used, no. (%)
Duration of RA,
mean ? SD years‡
59 ? 12
2.2 ? 1.3
55 ? 12
2.4 ? 1.6
53 ? 12
2.8 ? 2.2
10 ? 1
10 ? 1
10 ? 1
* RF ? rheumatoid factor; DMARDs ? disease-modifying antirheu-
matic drugs (see Table 1 for other definitions).
† For patients with AA genotype versus patients with GG or GA
genotypes considered together.
‡ From disease onset to radiographic analysis.
SDF-1 3?-UTR 801 genotypes*
Radiographic progression in RA patients with different
Latest radiographic score
Annual rate of progression
40 ? 47
41 ? 43
80 ? 66†
46 ? 42
39 ? 30
39 ? 30
60 ? 40‡
41 ? 31
83 ? 73
80 ? 72
139 ? 105§
87 ? 77
4.4 ? 4.8
4.1 ? 4.3
7.8 ? 6.4¶
3.8 ? 3
3.9 ? 3.2
5.9 ? 4#
8.3 ? 7.5
8.1 ? 7.2
13.7 ? 10.4**
* Values are the mean ? SD scores or changes in scores. See Table 1
† P ? 0.012 versus patients with GG or GA genotypes considered
‡ P ? 0.030 versus patients with GG or GA genotypes considered
§ P ? 0.013 versus patients with GG or GA genotypes considered
¶ P ? 0.018 versus patients with GG or GA genotypes considered
# P ? 0.042 versus patients with GG or GA genotypes considered
** P ? 0.020 versus patients with GG or GA genotypes considered
CONCISE COMMUNICATIONS 355
FJ, et al. Synoviocyte-derived CXCL12 is displayed on endothe-
lium and induces angiogenesis in rheumatoid arthritis. J Immunol
5. Kanbe K, Takagishi K, Chen Q. Stimulation of matrix metallopro-
tease 3 release from human chondrocytes by the interaction of
stromal cell–derived factor 1 and CXC chemokine receptor 4.
Arthritis Rheum 2002;46:130–7.
6. Mazzetti I, Magagnoli G, Paoletti S, Uguccioni M, Olivotto E,
Vitellozzi R, et al. A role for chemokines in the induction of
chondrocyte phenotype modulation. Arthritis Rheum 2004;50:
7. Yu X, Huang Y, Collin-Osdoby P, Osdoby P. Stromal cell-derived
factor-1 (SDF-1) recruits osteoclast precursors by inducing che-
motaxis, matrix metalloproteinase-9 (MMP-9) activity, and colla-
gen transmigration. J Bone Miner Res 2003;18:1404–18.
8. Grassi F, Cristino S, Toneguzzi S, Piacentini A, Facchini A,
Lisignoli G. CXCL12 chemokine up-regulates bone resorption and
MMP-9 release by human osteoclasts: CXCL12 levels are in-
creased in synovial and bone tissue of rheumatoid arthritis pa-
tients. J Cell Physiol 2004;199:244–51.
9. Matthys P, Hatse S, Vermeire K, Wuyts A, Bridger G, Henson
GW, et al. AMD3100, a potent and specific antagonist of the
stromal cell-derived factor-I chemokine receptor CXCR4, inhibits
autoimmune joint inflammation in IFN-? receptor-deficient mice.
J Immunol 2001;167:4686–92.
10. Canete JD, Pablos JL, Sanmarti R, Mallofre C, Marsal S, Maymo
J, et al. Antiangiogenic effects of anti–tumor necrosis factor ?
therapy with infliximab in psoriatic arthritis. Arthritis Rheum
11. Winkler C, Modi W, Smith MW, Nelson GW, Wu X, Carrington
M, et al. Genetic restriction of AIDS pathogenesis by an SDF-1
chemokine gene variant: ALIVE Study, Hemophilia Growth and
Development Study (HGDS), Multicenter AIDS Cohort Study
(MACS), Multicenter Hemophilia Cohort Study (MHCS), San
Francisco City Cohort (SFCC). Science 1998;279:389–93.
12. Dubois-Laforgue D, Hendel H, Caillat-Zucman S, Zagury JF,
Winkler C, Boitard C, et al. A common stromal cell-derived
factor-1 chemokine gene variant is associated with the early onset
of type 1 diabetes. Diabetes 2001;50:1211–3.
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;31:315–24.
14. Van der Heijde DMFM. How to read radiographs according to the
Sharp/Van der Heijde method. J Rheumatol 1999;26:743–5.
15. Soriano A, Martinez C, Garcia F, Plana M, Palou E, Lejeune M,
et al. Plasma stromal cell-derived factor (SDF)-1 levels, SDF1-3?A
genotype, and expression of CXCR4 on T lymphocytes: their
impact on resistance to human immunodeficiency virus type 1
infection and its progression. J Infect Dis 2002;186:922–31.
16. Rueda P, Garcia-Moruja C, Torres MC, Alcami J, Caruz A. Partial
resistance to HIV-1 disease progression induced by SDF1 3?UTR
A801G polymorphism: a molecular explanation. Proceedings of
the 13th International Symposium in HIV and Emerging Infec-
tious Diseases; 2004; Toulon, France.
Evidence for two phenotypes in the metabolism of
methotrexate to 7-hydroxymethotrexate in patients
with rheumatoid arthritis
We have observed that patients with rheumatoid ar-
thritis (RA) vary greatly in their capacity to metabolize
methotrexate (MTX) to 7-hydroxymethotrexate (7-OH-MTX)
(1). Here we present evidence of two phenotypes for this
metabolism, suggesting two underlying genotypes for aldehyde
oxidase (AO; EC18.104.22.168), which catalyzes the oxidation of
MTX to 7-OH-MTX (2–5).
The RA patients in the present study participated in an
Institutional Review Board–approved protocol described pre-
viously (1). Each patient had received a stable weekly oral dose
of MTX (7.5–22.5 mg) for 2 months before we started collect-
ing urine after each MTX dose. The volume of the urine
sample was measured, and MTX and 7-OH-MTX concentra-
tions were determined using the high-performance liquid
chromatography method previously described (1). Urine was
collected from 29 patients in the following time periods after
the MTX dose: 0–2, 2–5, 5–8, 8–12, 12–17, 17–24, 24–36,
36–48, and 48–72 hours.
The following methods were employed to test for
phenotypes for the ability to metabolize MTX to 7-OH-MTX.
First, the distribution of the percentage of the total MTX dose
excreted in the urine over 72 hours as 7-OH-MTX is presented
in Figure 1. The distribution was tested for normality using
?0–0.3, ?0.3–0.6, ?0.6–1.0, and ?1.0–? SD intervals and the
chi-square test (6).
Second, the in vivo maximum velocity of 7-OH-MTX
production (i.e., Vmax) and the in vivo Michaelis-Menten
constant (Km) for this metabolism were estimated using non-
linear regression of data sets fit to the Michaelis-Menten
equation (EZ-fit Program; E. I. DuPont de Nemours, Wil-
mington, DE). These data sets were generated by assuming
that the total MTX plus 7-OH-MTX excreted in 72 hours was
the body pool of MTX available for oxidation to 7-OH-MTX at
a time shortly after (i.e., 1 or 2 hours) ingestion of the drug.
The rate of formation of 7-OH-MTX was the amount of
7-OH-MTX found in a urine collection divided by the number
of hours in that collection. This was the rate of formation of
7-OH-MTX from MTX from an average body pool of MTX
remaining. Thus, the rate of 7-OH-MTX formation and the
average body pool of MTX formed data sets that were fit to the
Michaelis-Menten equation. Data sets from the 6 highest
metabolizers of MTX (mean ? SD 11.9 ? 0.7% of the dose)
and the 6 lowest metabolizers (mean ? SD 2.8 ? 1.1% of the
dose) were used and combined to produce one set for the highest
metabolizers and another set for the lowest metabolizers.
The distribution of percentage excretion of the dose of
MTX as 7-OH-MTX is presented in Figure 1. The mean ? SD
Figure 1. Distribution of the percentage of the dose of methotrexate
excreted in the urine over 72 hours as 7-hydroxymethotrexate (7-OH-
MTX) in 29 patients with rheumatoid arthritis.
356 CONCISE COMMUNICATIONS
was 7.19 ? 3.37%, the median value was 6.3%, and the range
was 0.94–13.2%. Two modes are apparent in Figure 1, and the
distribution failed the test for normality (?2? 11.4, 4 degrees
of freedom, P ? 0.025). The percentage of the MTX dose
excreted as 7-OH-MTX was not significantly correlated with
the MTX dose level or the percentage of the MTX dose
excreted as MTX plus 7-OH-MTX.
There was no difference in the mean ? SD percentage
of total MTX dose excreted as MTX plus 7-OH-MTX in urine
between the 6 lowest and the 6 highest 7-OH-MTX producers
(66.7 ? 18.1% and 69.7 ? 10.3%, respectively). There was also
no difference in mean ? SD (median) MTX dose between the
lowest and highest 7-OH-MTX producers (12.9 ? 4.6 mg [10
mg] and 11.2 ? 3.8 mg [10 mg], respectively). Twenty-three
and 28 data sets of rate of 7-OH-MTX production and average
body pool of MTX were fit to the Michaelis-Menten equation
for the low and high 7-OH-MTX producers, respectively. The
mean ? SEM Vmax and Kmwere 24 ? 1 ?g/hour and 510 ?
80 ?g versus 73 ? 10 ?g/hour and 2,300 ? 500 ?g, respectively,
for the low versus the high 7-OH-MTX producers. Both Vmax
and Kmwere significantly different (P ? 0.005 by t-test).
The mean percentage of the MTX dose excreted in
urine as 7-OH-MTX was somewhat higher than the 2–5%
reported previously in low oral dose MTX–treated psoriasis
and RA patients (7–9). This was probably due to a longer (i.e.,
72 hours) collection time. Marked interpatient variability in
this percentage (i.e., a 14-fold difference [13.2 ? 0.94]) is
reported here, and a factor of 4–5 has been reported by others
(7–9). Analysis of MTX and 7-OH-MTX in bone marrow of
leukemia patients receiving low-dose MTX indicates an enor-
mous interpatient variability in this metabolism (10). Our data
indicate that the variation in the ability to metabolize MTX to
7-OH-MTX is caused in part by two phenotypes (Figure 1).
In vitro data measuring human liver AO activities,
substrate specificity, and kinetics also suggest more than one
phenotype and genotype. Human livers are reported to vary
40–50-fold in their AO-catalyzed oxidation of benzaldehyde
and N-methyl-nicotinamide in vitro (11,12). Kmand Vmax
values for AO varied 3–4-fold (13). The lability of AO and of
the in vitro assays is a source of concern regarding the above
results, since in vivo conditions and the cofactor levels for this
iron-, riboflavin-, and molybdopterin-dependent enzyme are
difficult to reproduce.
We think it is valid to analyze patient data with enzyme
kinetic parameters. Polyglutamates of MTX are not effective
substrates for AO (3). Both MTX and 7-OH-MTX can readily
cross cellular membranes; therefore, urinary excretion of
7-OH-MTX should be proportional to its in vivo formation,
and urinary excretion of MTX will deplete the amount of MTX
remaining (i.e., the body pool of metabolizable MTX). Thus,
the calculated Vmax and Kmparameters are likely to be
proportional to those for the actual in vivo metabolism of
The fact that both Kmand Vmax parameters are
different suggests that there are two AO genotypes. Only one
human liver enzyme has been identified; however, multiple
transcription-initiator sites are found in the human liver gene
(4,5). There are 199 single-nucleotide polymorphisms (SNPs)
in the human AO gene listed in the National Center for
Biotechnology Information SNP database, but the functional
consequences of these SNP alleles/haplotypes remain un-
known. Alternate explanations for our data would include the
following: 1) posttranslational modification of one gene prod-
uct, thereby producing two enzymes; 2) enzymes from differ-
ent organs (other than the liver) expressed to a different
extent; and 3) low producers of 7-OH-MTX having very high
levels of folyl-polyglutamate synthetase, which could prevent
the formation of 7-OH-MTX by rapidly producing polygluta-
mates of MTX and prevent the excretion of 7-OH-MTX by
sequestering it as a polyglutamate metabolite. The last possi-
bility could also result from two genotypes.
The extent of the metabolism of MTX to 7-OH-MTX
could play an important role in its efficacy and toxicity. In rat
adjuvant-induced arthritis, 7-OH-MTX is less efficacious than
Supported by NIH grant P60-AR-4805 and by the University of
Alabama at Birmingham Multidisciplinary Clinical Research Center.
Joseph E. Baggott, PhD
S. Louis Bridges, Jr., MD, PhD
Sarah L. Morgan, MD, MS, RD
University of Alabama at Birmingham
1. Fox RI, Morgan SL, Smith HT, Robbins BA, Choc MG, Baggott
JE. Combined oral cyclosporin and methotrexate therapy in
patients with rheumatoid arthritis elevates methotrexate levels and
reduces 7-hydroxymethotrexate levels when compared with meth-
otrexate alone. Rheumatology (Oxford) 2003;42:989–94.
2. Johns DG, Iannotti AT, Sartorelli AC, Booth BA, Bertino JR.
Enzymic oxidation of methotrexate and aminopterin. Life Sci
3. Fabre G, Seither R, Goldman ID. Hydroxylation of 4-amino-
antifolates by partially purified aldehyde oxidase from rabbit liver.
Biochem Pharmacol 1986;35:1325–30.
4. Beedham C, Critchley DJ, Rance DJ. Substrate specificity of
human liver aldehyde oxidase toward substituted quinazolines and
phthalazines: a comparison with hepatic enzyme from guinea pig,
rabbit, and baboon. Arch Biochem Biophys 1995;319:481–90.
5. Terao M, Kurosaki M, Demontis S, Zanotta S, Garattini E.
Isolation and characterization of the human aldehyde oxidase
gene: conservation of intron/exon boundaries with the xanthine
oxidoreductase gene indicates a common origin. Biochem J 1998;
6. Pollard JH. A handbook of numerical and statistical techniques
with examples mainly from the life sciences. Cambridge: Cam-
bridge University Press; 1977.
7. Seideman P, Beck O, Eksborg S, Wennberg M. The pharmacoki-
netics of methotrexate and its 7-hydroxy metabolite in patients
with rheumatoid arthritis. Br J Clin Pharmacol 1993;35:409–12.
8. Chladek J, Martinkova J, Simkova M, Vaneckova J, Koudelkova
V, Nozickova M. Pharmacokinetics of low doses of methotrexate
in patients with psoriasis over the early period of treatment. Eur
J Clin Pharmacol 1998;53:437–44.
9. Chladek J, Grim J, Martinkova K, Simkova M, Vaniekova J,
Koudelkova V, et al. Pharmacokinetics and pharmacodynamics of
low-dose methotrexate in the treatment of psoriasis. Br J Clin
10. Sonneveld P, Schultz FW, Nooter K, Hahlen K. Pharmacokinetics
of methotrexate and 7-hydroxy-methotrexate in plasma and bone
marrow of children receiving low-dose oral methotrexate. Cancer
Chemother Pharmacol 1986;18:111–6.
11. Rodrigues AD. Comparison of levels of aldehyde oxidase with
cytochrome P450 activities in human liver in vitro. Biochem
12. Sugihara K, Kitamura S, Tatsumi K, Asahara T, Dohi K. Differ-
ences in aldehyde oxidase activity in cytosolic preparations of
human and monkey liver. Biochem Mol Biol Int 1997;41:1153–60.
13. Al-Salmy HS. Individual variation in hepatic aldehyde oxidase
activity. IUBMB Life 2001;51:249–53.
14. Baggott JE, Morgan SL, Koopman WJ. The effect of methotrexate
and 7-hydroxymethotrexate on rat adjuvant arthritis and on uri-
nary aminoimidazole carboxamide excretion. Arthritis Rheum
Association of R602W in a protein tyrosine
phosphatase gene with a high risk of rheumatoid
arthritis in a British population: evidence for an early
onset/disease severity effect
The concept of a strong genetic component to suscep-
tibility to rheumatoid arthritis (RA) is well established, and the
HLA-DRBI locus is estimated to account for ?30% of this.
However, despite several genome-wide linkage studies and
numerous association studies involving positional and/or func-
tional candidate genes, there has been limited success in the
identification of further RA susceptibility loci. Failure to
replicate reported associations in complex diseases is common-
place, as recently demonstrated by a lack of replication in a
well-powered study (1) of a published strong association
between RA and PADI4 mutations (2). Independent study
replication is therefore vital, and enables both the further
refinement of disease risk estimates for associated polymor-
phisms (as demonstrated by HLA-DRB1 locus studies in RA)
and the consideration of interactions with other loci and with
clinical and epidemiologic factors, the value of which has
recently been demonstrated in Crohn’s disease (3).
The pathology of RA is characterized by synovial
inflammation with infiltration by CD4? T cells, B cells, and
macrophages. Therefore, genes that encode proteins with
functions that influence the cellular immune response, and
also lie within previously identified regions of linkage, provide
excellent candidate genes for disease susceptibility. PTPN22
resides in chromosomal region 1p13, which has been previ-
ously shown to be linked to RA in North American families
(4), and encodes a protein tyrosine phosphatase which inter-
acts with the negative regulatory kinase Csk to inhibit T cell
signaling and activation. Bottini and colleagues (5) demon-
strated an association between the rare T allele of a single-
nucleotide polymorphism (SNP), (rs2476601, 1858C/T), and
type 1 diabetes in North American and Sardinian populations.
This SNP, which lies within the first proline-rich domain of
PTPN22, results in the substitution of tryptophan for arginine
at codon 620 (R620W) of PTPN22, and has been shown to
affect the binding of PTPN22 to Csk (5,6). It has been
postulated that the reduced interaction of the variant phos-
phatase with Csk may result in failure of down-regulation of T
cell activation, which in turn may account for the recently
reported strong association of this missense SNP with RA in a
North American population (6).
The aim of this study was to replicate and extend the
reported association of the 1858C/T SNP with RA in a group
of 302 UK Caucasian RA cases and 374 healthy UK Caucasian
controls. Ethical approval was obtained for the recruitment of
patients with RA from clinics at Guy’s & St Thomas’ Hospital,
and Lewisham Hospital in London. These cases satisfied the
1987 American College of Rheumatology diagnostic criteria
for RA (7), and were recruited consecutively. In total, 234
(77%) of cases were female, 79% were rheumatoid factor (RF)
positive, and 87% had erosive disease. Written informed
consent for participation was obtained, and genomic DNA was
extracted from samples of whole blood. Controls were ob-
tained from the European Collection of Animal Cell Cultures
(Wiltshire, UK) and the Clinical Transplantation Laboratory
at Guy’s Hospital.
Cases and controls were genotyped for rs2476601 using
a TaqMan Assay By Design (Applied Biosystems, Warrington,
Cheshire, UK). Primer and probe sequences were as follows:
forward primer 5?-CCAGCTTCCTCAACCACAATAAATG-
3?, reverse primer 5?-CAACTGCTCCAAGGATAGA-
TGATGA-3?, probe for C allele 5?-FAM TCAGG-
TGTCCGTACAGG-3?, probe for T allele 5?-VIC TCAGGT-
GTCCATACAGG-3?. This assay was validated using a previ-
ously published method (5). HLA class II typing using
sequence-specific primers revealed that 78% of cases carried 1
or more copies of the HLA-DRB1 shared epitope (8), which
was defined as the presence of 1 or more of the DRB1 alleles
*0101, *0102, *0401, *0404, *0405, *0408, *1001, and *1402.
Genotypes were in Hardy-Weinberg equilibrium in
cases and controls. The distribution of genotypes in cases and
controls is shown in Table 1. The frequency of the minor allele
(T) in cases was 15.9%, which was significantly higher than the
8.4% frequency in controls (P ? 0.00003 by chi-square test).
PTPN22 C1858T SNP case–control analysis*
OR 95% CIOR95% CI
Controls (n ? 374)
RA cases (n ? 302)
Negative (n ? 58)
Positive (n ? 223)
Absent (n ? 37)
Present (n ? 249)
15.90.000031.70 1.15–2.48 17.173.79–77.79
* SNP ? single-nucleotide polymorphism; OR ? odds ratio; 95% CI ? 95% confidence interval; RA ? rheumatoid arthritis.
† Clinical data were not available for some cases.
These allele frequencies are similar to those previously re-
ported in North American RA cases (14.8%) and controls
(8.7%) (6). The odds ratios for CT and TT genotypes were 1.7
(95% confidence interval [95% CI] 1.2–2.5) and 17.2 (95% CI:
3.8–77.8) respectively. The disease risk for homozygous TT
individuals observed here is much higher than the ?2-fold risk
seen in North American RA cases. Consistent with the previ-
ous study, genotype frequencies were similar in HLA-DRB1
shared epitope–positive and shared epitope–negative cases,
suggesting that the PTPN22 gene acts independently from the
The mean age at disease onset for each genotypic
subgroup of cases was calculated (Table 2). A negative corre-
lation was observed between mean age at onset and number of
T alleles present in RA cases. Survival curves (years until RA
onset) for each genotypic subgroup were calculated using
Kaplan-Meier survival analysis (implemented using Splus ver-
sion 6.0) and compared using a Mantel-Haenszel test. A
significantly earlier median age at onset was observed in TT
homozygous cases (42.2 years) than in CC homozygous cases
(49.7 years) (P ? 0.003). The median age at disease onset in
heterozygous cases (45 years) was not significantly different
from that in homozygous CC cases (P ? 0.25). Stratification of
cases by clinical factors (Table 1) revealed that the frequency
of allele T was higher in RF-positive cases (18.2%) than
RF-negative cases (11.2%), and higher in cases with erosive
disease (17.5%) compared with those who had nonerosive
disease (10.8%), although these differences were not statisti-
cally significant. These data are not adjusted for disease
duration, but only 7 patients (2%) had disease for ?1 year (and
therefore might be subject to misclassification with regard to
erosions). Disease duration would therefore be unlikely to
substantially affect this observation. The frequency of the
PTPN22 minor allele in RF-positive cases is consistent with
that seen in RF-positive cases in the previous study (6).
The protein tyrosine phosphatase encoded by PTPN22
is expressed in hematopoietic tissues and all subtypes of
human peripheral blood mononuclear cells including mono-
cytes, neutrophils and natural killer cells, as well as T cells and
B cells. It is not yet known whether the genetic variant
investigated here affects function in all of these cell types, and
if so, how this contributes to disease pathogenesis in RA. The
association between PTPN22 and both RA and type 1 diabetes
is consistent with the hypothesis of shared susceptibility loci for
autoimmune diseases (9). It remains to be seen whether the
disease mechanism underlying the association of this polymor-
phism with RA will also explain the association with type 1
diabetes and potentially other autoimmune diseases.
In summary, we have replicated and extended the
recently reported association of a missense SNP in codon 620
of the PTPN22 gene with RA in an independent group of cases
from a UK Caucasian population. The genotype-specific dis-
ease risks estimated from our cohort suggest a gene-dosage
model associated with this mutation. Our risk estimates are
similar to those associated with CARD15 mutations in Crohn’s
disease (3,10) but contrast with the previous estimated 2-fold
risk for rare homozygotes conferred by PTPN22 in RA (6).
This may be due in part to the low frequency of controls
homozygous for the rare allele in the British cohort. The large
confidence interval around this estimate reflects our sample
size. However, the nonoverlapping confidence intervals for the
risk estimates demonstrate a much greater risk of RA associ-
ated with the presence of 2 risk alleles (odds ratio ? 17) than
a single risk allele, although formal testing of genetic models
shows that the gene-dosage model is a better fit than a
recessive mode of inheritance model. Importantly, we have
also identified a significant association between earlier age at
disease onset and the presence of PTPN22 mutations, which
occur at high frequency in both patients with erosive disease
and RF-positive patients. Further association studies of large
cohorts and/or meta-analyses of previous sample groups will
enable refinement of the true risk of disease associated with
specific genotypes, and elucidate the relationship between
PTPN22 mutations and the clinical and laboratory features of
Supported by the Arthritis Research Campaign and NHS Re-
search and Development Support Funding for Guy’s and St. Thomas’ and
Lewisham NHS Trusts. The authors would like to thank the patients who
participated in this study. The authors would also like to thank Dr. E.
Choy, Dr. T. Gibson, Dr. V. Hajela, Dr. B. Kirkham, Professor A.
MacGregor, Professor G. Panayi, Professor T. Spector, and Dr. G. Yanni
for allowing access to their patients, and Mrs. T. Hateley for assistance
with data entry.
Sophia Steer, MB, MSc, MRCP
Bhaneeta Lad, BSc
Janet A. Grumley, B. Nur (Hon)
Gabrielle H. Kingsley, MB, PhD, FRCP
Sheila A. Fisher, PhD
Guy’s, King’s and St Thomas’ School of Medicine
1. Barton A, Bowes J, Eyres S, Spreckley K, Hinks A, John S, et al.
A functional haplotype of the PADI4 gene associated with rheu-
matoid arthritis in a Japanese population is not associated in a
United Kingdom population. Arthritis Rheum 2004;50:1117–21.
2. Suzuki A, Yamada R, Chang X, Tokuhiro S, Sawada T, Suzuki M,
et al. Functional haplotypes of PADI4, encoding citrullinating
enzyme peptidylarginine deiminase 4, are associated with rheuma-
toid arthritis. Nat Genet 2003;34:395–402.
3. Cuthbert AP, Fisher SA, Mirza MM, King K, Hampe J, Croucher
PJ, et al. The contribution of NOD2 gene mutations to the risk and
site of disease in inflammatory bowel disease. Gastroenterology
4. Jawaheer D, Seldin MF, Amos CI, Chen WV, Shigeta R, Etzel C,
et al. Screening the genome for rheumatoid arthritis susceptibility
genes: a replication study and combined analysis of 512 multicase
families. Arthritis Rheum 2003;48:906–16.
5. Bottini N, Musumeci L, Alonso A, Rohmouni S, Nika K, Ros-
tamkhani M, et al. A functional variant of lymphoid tyrosine phos-
phatase is associated with type I diabetes. Nat Genet 2004;36:337–8.
6. Begovich AB, Carlton VE, Honigberg LA, Schrodi SJ, Chokkal-
ingam. AP, Alexander HC, et al. A missense single-nucleotide
Comparison of age at disease onset between genotype
CC CT TT
Mean ? SD
49.2 ? 13.2
45.9 ? 14.8
40.6 ? 11.1
* P ? 0.25 versus CC group.
† P ? 0.003 versus CC group.
polymorphism in a gene encoding a protein tyrosine phosphatase
(PTPN22) is associated with rheumatoid arthritis. Am J Hum
7. 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;31:315–24.
8. Gregersen PK, Silver J, Winchester RJ. The shared epitope
hypothesis: an approach to understanding the molecular genetics
of susceptibility to rheumatoid arthritis. Arthritis Rheum 1987;30:
9. Becker KG, Simon RM, Bailey-Wilson JE, Freidlin B, Biddison
WE, McFarland HF, et al. Clustering of non-major histocompat-
ibility complex susceptibility candidate loci in human autoimmune
diseases. Proc Nat Acad Sci U S A 1998;95:9979–84.
10. Hugot JP, Chamaillard M, Zouali J, Lesage S, Cezard JP, Belaiche
J, et al. Association of NOD2 leucine-rich repeat variants with
susceptibility to Crohn’s disease. Nature 2001;411:599–603.