close

Вход

Забыли?

вход по аккаунту

?

Lack of association of the HLADRB1 shared epitope with rheumatoid nodulesAn individual patient data meta-analysis of 3272 Caucasian patients with rheumatoid arthritis.

код для вставкиСкачать
ARTHRITIS & RHEUMATISM
Vol. 50, No. 3, March 2004, pp 753–762
DOI 10.1002/art.20119
© 2004, American College of Rheumatology
Lack of Association of the HLA–DRB1
Shared Epitope With Rheumatoid Nodules
An Individual Patient Data Meta-Analysis of
3,272 Caucasian Patients With Rheumatoid Arthritis
Jennifer D. Gorman,1 Eve David-Vaudey,2 Madhukar Pai,3 Raymond F. Lum,4
and Lindsey A. Criswell4
Objective. The objective of this individual patient
data (IPD) meta-analysis was to examine the relationship of rheumatoid nodules to the HLA–DRB1 shared
epitope (SE) and to individual SE genotypes.
Methods. English-language studies that enrolled
adult non-Hispanic Caucasian patients with rheumatoid arthritis (RA) were identified by searches of Medline and Embase, and by manual searches of medical
journals. All authors were contacted for IPD. Metaanalysis was performed to assess the association of SE
presence, dose, and genotype with rheumatoid nodules.
Meta-analyses adjusted for disease duration and cumulative meta-analyses were also performed to assess the
influence of RA duration and year of study publication
on the results.
Results. A total of 24 studies and 3,272 patients
were available for analysis. IPD were obtained for 22 of
the studies. There was a nonsignificant association
between the presence of the SE (i.e., 1 or 2 alleles versus
0 alleles) and rheumatoid nodules (summary odds ratio
[OR] 1.3, 95% confidence interval [95% CI] 0.97–1.6).
Analysis by SE genotype, however, demonstrated a weak
relationship with inheritance of a single DRB1*0401 SE
allele (OR 1.4, 95% CI 1.1–1.8). No other genotypes
achieved statistical significance in the adjusted or unadjusted analyses.
Conclusion. The presence of the HLA–DRB1 SE
does not appear to significantly increase the risk of
rheumatoid nodules among Caucasian patients with
RA. Analysis by DRB1 SE genotype was uninformative,
suggesting only a potential (and at most modest) role of
the DRB1*0401 SE allele. Results from this IPD metaanalysis implicate other genetic, stochastic, and/or environmental factors in the susceptibility to rheumatoid
nodules.
Marked heterogeneity exists in the clinical presentation of rheumatoid arthritis (RA) between individual patients. This is particularly true for development of
extraarticular manifestations, typically defined as manifestations of disease distinct from the erosive and deforming joint process. One of the most commonly encountered extraarticular manifestations is rheumatoid
nodules, which are estimated to affect ⬃20% of patients
with RA (1). The presence of extraarticular manifestations, including nodules, has long been considered a sign
of greater RA severity, as supported by results of recent
studies demonstrating 2–4-fold increased mortality in
patients with these manifestations (2,3).
Genetic variation is believed to be important in
determining both the susceptibility to and severity of
Supported by the American College of Rheumatology, the
NIH (grants AR-07304, AR-20684, and AR-02175, and the Fogarty
AIDS International Training Program [grant 1-D43-TW00003]), the
Arthritis Foundation, and the Rosalind Russell Medical Research
Center for Arthritis.
1
Jennifer D. Gorman, MD, MPH: University of California,
San Francisco, and School of Public Health, University of California,
Berkeley; 2Eve David-Vaudey, MD: University of Paul Sabatier,
Toulouse, France; 3Madhukar Pai, MD: School of Public Health,
University of California, Berkeley; 4Raymond F. Lum, MPH, Lindsey
A. Criswell, MD, MPH: University of California, San Francisco.
Address correspondence and reprint requests to Lindsey A.
Criswell, MD, MPH, Division of Rheumatology, University of California, San Francisco, 374 Parnassus Avenue, Box 0500, San Francisco,
CA 94143-0500. E-mail: lac@itsa.ucsf.edu.
Submitted for publication August 8, 2003; accepted in revised
form December 11, 2003.
753
754
GORMAN ET AL
RA, and, to date, the most illuminating genetic theory
has been that of the shared epitope (SE) (4). The SE
hypothesis was proposed to explain susceptibility to
RA, based on the observation that HLA–DR specificities associated with disease share a similar amino acid
sequence in the third hypervariable region of the DRB1
molecule. Although preliminary research on the SE
supported an association with RA susceptibility, results
of subsequent investigations suggested that its primary
role may be in disease severity. The most striking results
were generated from one of the earliest analyses,
demonstrating a marked association of nodules and
other extraarticular manifestations with SE alleles in a
highly selected group of patients with RA (5). Since
then, numerous studies have pursued clarification of
this relationship, although subsequent findings have
been much less striking, with many investigations
failing to demonstrate a significant association with
outcome (6–8).
The difficulty in interpreting the relationship of
the SE with RA severity is likely attributable, at least in
part, to differences in the populations studied, such as
ethnicity, clinical characteristics, and the particular SE
allele(s) inherited, as well as the relatively small size of
most studies. The combination of the relative infrequency of nodules and the small size of individual
studies motivated us to perform a meta-analysis. In
order to investigate the relationship more thoroughly
than is possible using traditional meta-analytic techniques (i.e., limited to summary data from published
articles), we obtained individual patient data (IPD) from
study authors. IPD meta-analysis has several advantages
over the traditional approach, including the ability to
incorporate unpublished and/or updated data, evaluate
and correct raw data, perform additional analyses of
subgroups, or adjust for potential confounding factors
(9). Therefore, our IPD meta-analysis allowed us the
unique opportunity to examine the role of specific SE
genotypes in the development of rheumatoid nodules
and to investigate the influence of covariates, such as
disease duration and sex.
MATERIALS AND METHODS
Inclusion criteria. Studies eligible for inclusion in this
meta-analysis were those in which HLA–DRB1 alleles in RA
patients were investigated, and in which information regarding
extraarticular manifestations was reported. The HLA–DRB1
alleles defined as encoding the SE were as follows: HLA–
DRB1*0101, *0102, and *0104 (DR1); HLA–DRB1*0401,
*0404, *0405, *0408, *0409, *0410, *0413, *0416, *0419, and
*0421 (DR4); HLA-DRB1*1001 (DR10); and HLA–
DRB1*1402 and *1406 (DR14) (10). The frequency of HLA–
DRB1 alleles varies according to ethnic and racial background,
with some alleles being extremely rare; therefore, articles were
not required to identify all 16 SE alleles to be eligible for
inclusion.
Studies selected were those that evaluated adult, nonHispanic, Caucasian patients and assessed for the presence of
rheumatoid nodules. Non-Caucasian and Hispanic patients
were excluded from this analysis in order to evaluate a more
genetically homogenous population and due to the limited
number of these studies available within the published literature. Only studies or letters written in English were included in
our analysis.
Search strategy. Three overlapping search strategies
were used to identify appropriate articles published between
January 1987 and June 1999. We searched Medline with the
truncated keywords of rheumatoid arthritis and 1 or more of
the following: human leukocyte antigen, gene, prognosis, or
severity. Embase was searched with the exploded terms rheumatoid arthritis and 1 or more of the following: histocompatibility, HLA system, epitope, etiology, or genetics. The database search terms were designed to be exceedingly sensitive in
order to identify all studies involving HLA–DRB1 genotyping
in which there was any reference to RA severity. In addition,
an updated PubMed search was performed to identify articles
published between June 1999 and March 2003.
Hand searches of 6 journals (Annals of the Rheumatic
Diseases, Arthritis & Rheumatism, British Journal of Rheumatology, Journal of Rheumatology, Scandinavian Journal of Rheumatology, and Tissue Antigens) for articles (January 1987
through June 1999) were also performed. We selected the start
of the time period for article identification based on the
introduction of molecular DRB1 typing in 1987 (11).
A single investigator (JDG) performed the literature
searches and reviewed abstracts identified by the search.
Training on database and hand searching was provided by
formal instruction prior to the start of the study. Upon
identification of articles appropriate for study inclusion, letters
were mailed or electronically sent to the corresponding author
to request missing information and/or IPD. If the same patients were described in more than 1 article (i.e., duplicate
data) and IPD were not available, the publication that presented the most complete recent information was selected for
inclusion in the analysis. One investigator (JDG) performed
the data extraction from studies without IPD. Because the IPD
were provided without personal identifiers, this study was
certified as exempt from Human Subjects Review by the
University of California, San Francisco Committee on Human
Research.
The influence of individual study quality on the results
of meta-analyses has been well described, particularly for
meta-analyses of clinical trials (12). Although assessment of
study quality is considered to be an important component of
systematic reviews and meta-analyses, it has been demonstrated that use of such scoring methods can be problematic
(13,14) and may not accurately assess the quality measures of
interest (15,16). Moreover, in practice, most systematic reviews
of published studies assess the quality of data reporting, which
does not always correspond precisely with the quality of the
study conducted. Because we obtained IPD, we did not conduct a formal assessment of study reporting. However, we used
Women only
RA after vaccination
Cross-sectional
Cross-sectional
Prospective cohort
Case series
Cross-sectional
Case–control
Multiplex§
87
120
109
100
Case–control
Case–control
Case–control
Cross-sectional
3,272
136.3 ⫾ 90.3
111
299
171
9
77
88
95
83
45
239
131
Cross-sectional
Case–control
Case–control
Cross-sectional
Cross-sectional
128
403
69
36
200
216
218
66
172
Case–control
Prospective cohort
Cross-sectional
Cross-sectional
Case–control
Cross-sectional
Cross-sectional
Retrospective cohort
Prospective cohort
RF⫹, erosive disease
Early RA
RF⫹
Multiplex§
Multiplex§
Erosive disease
Early RA
Early RA
Early RA
Study design
No. of
patients†
French
French
French
French
European
American
American
Canadian
European, Australian
S. and N. French
Turkish
Greek
Greek
S.
S.
S.
S.
Swiss
German
British
British
N. French
British
N. French
British
British
Swedish
Swedish
Ethnicity‡
59.9 ⫾ 4.5
–
63
63
–
–
–
–
57
–
–
61
50
–
65
–
–
–
59
56
59
–
54
NA
–
Age,
mean
years
10.8 ⫾ 6.6
–
16
21
3
18
8
12
7
11
9
10
0.6
–
16
–
1
0.9
12
22
13
–
14
17
–
Disease
duration,
mean years
73.6 ⫾ 10.4
73
56
100
56
66
85
80
82
87
80
74
79
70
74
–
66
67
80
83
72
–
84
77
64
% female
75.6 ⫾ 13.5
90
88
78
75
92
72
100
84
78
62
85
69
73
73
100
72
44
75
95
85
–
83
57
70
% RF⫹
* Individual patient data were not available for Hall et al (30) and Seidl et al (33). SE ⫽ shared epitope; RF ⫽ rheumatoid factor; NA ⫽ not available; early rheumatoid arthritis
(RA) ⫽ disease duration ⬍2 years; N ⫽ northern; S ⫽ southern.
† Available for at least 1 analysis.
‡ American, European, and Australian populations were composed entirely of Caucasians.
§ Family history of RA; only 1 family member was included in the meta-analysis.
Total
Mean ⫾ SD
Scandinavian
Turesson et al, 2000 (8)
Eberhardt and Fex, 1998 (24)
Northern European
Vaidya et al, 2002 (25)
Thomson et al, 1999 (26)
Cortet et al, 1997 (27)
McDonagh et al, 1997 (28)
Perdriger et al, 1997 (29)
Hall et al, 1996 (30)
MacGregor et al, 1995 (31)
Central European
Bas et al, 2000 (32)
Seidl et al, 1999 (33)
Southern European
Hayem et al, 1999 (34)
Toussirot et al, 1999 (35)
Cantagrel et al, 1999 (36)
Benazet et al, 1995 (37)
Mediterranean
Saruhan-Direskeneli, 1998 (38)
Stavropoulos et al, 1997 (39)
Boki et al, 1993 (40)
Other/mixed Caucasian
Bali et al, 1999 (41)
Meyer et al, 1999 (42)
Criswell et al, 1998 (43)
Pope et al, 1998 (44)
Rowley et al, 1997 (45)
Combe et al, 1995 (46)
Patient selection
requirements
Characteristics of studies included in the meta-analysis of the association of the SE with rheumatoid nodules*
Author, year (ref.)
Table 1.
NO RELATIONSHIP OF RHEUMATOID NODULES TO THE HLA–DRB1 SE
755
756
GORMAN ET AL
Table 2.
Clinical and genetic characteristics of RA patients with or without nodules*
Characteristic
Nodular RA
(n ⫽ 1,092)
RA without nodules
(n ⫽ 2,180)
P
% male (n ⫽ 2,924)
Age, mean ⫾ SD years (n ⫽ 1,125)
Age at onset, mean ⫾ SD years (n ⫽ 2,898)
Disease duration, mean ⫾ SD years (n ⫽ 2,536)
% rheumatoid factor positive (n ⫽ 3,019)
% with erosive disease (n ⫽ 2,162)
30.2
59.6 ⫾ 13.0
45.4 ⫾ 14.4
14.2 ⫾ 11.2
88.2
89.0
24.1
58.9 ⫾ 14.1
49.0 ⫾ 15.0
9.1 ⫾ 10.0
70.0
66.5
⬍0.001
0.4
⬍0.001
⬍0.001
⬍0.001
⬍0.001
* P values were calculated by 2-tailed chi-square test or t-test. RA ⫽ rheumatoid arthritis; n ⫽ total
number of patients available for analysis.
study design as a quality variable (case–control/cross-sectional
versus cohort designs) and performed subgroup analyses to
evaluate its impact on effect measures. We also critically
appraised the quality of included studies to evaluate factors
such as bias in case selection or misclassification due to a
shorter duration of RA. One reviewer (JDG) evaluated the
impact of these quality variables.
Statistical analysis. The odds ratio (OR) for the
association between the SE and rheumatoid nodules was the
effect measure of interest. In all cases, patients with no SE
alleles served as the reference group for the computation of
ORs. For the meta-analysis, summary ORs were computed
using either fixed effects models (for data that did not demonstrate significant heterogeneity) or random effects models
(for data demonstrating significant heterogeneity) (17). The
Breslow-Day test for homogeneity was used to assess heterogeneity (18), and a significance level of P ⫽ 0.1 was established
to test consistency of the OR across studies (19).
Additional analyses were performed to account for
differences in disease duration between individuals. This approach required the use of multivariate logistic regression
analyses to calculate ORs for each of the individual studies
along with 95% confidence intervals (95% CIs), with the
presence of rheumatoid nodules as the outcome variable, the
SE (i.e., presence, dose, or allelic genotype) as the explanatory
variable, and disease duration as a covariate. IPD were required for this adjusted analysis because of the need for
information regarding disease duration for each patient. The
adjusted ORs after logistic regression were then pooled using
fixed effects or random effects meta-analyses. In addition, a
cumulative meta-analysis by publication year was performed to
assess the pattern of results over time. Sensitivity analyses
based on the definition of rheumatoid nodules by individual
studies were also performed.
Publication bias was assessed graphically by funnel plot
(20) and was statistically evaluated by the regression asymmetry test described by Egger et al (21), which has been demonstrated to be a powerful test for assessment of publication bias
(22). Specifically, the power to detect bias for a meta-analysis
of 20 studies with a control group event rate of 20% is
estimated to be ⬃50% for meta-analyses with moderate bias
and exceeds 95% for those with severe bias (22). All analyses
were performed using Stata statistical software, version 8.0
(23).
RESULTS
The database searches identified a total of 10,367
articles. Upon review, however, only 52 articles mentioned HLA–DRB1 and rheumatoid nodules in the title
or abstract. Eighteen of these articles were excluded
from our analysis: 12 because they evaluated nonCaucasian patients, and 6 that represented duplicate
publications. Authors were contacted for additional
data, and the authors of 22 of the 34 articles provided
IPD, with information regarding genotype and rheumatoid nodules in Caucasian patients. Two additional
articles from the original 34 did not include IPD, but
enough information was available within the published
articles regarding the SE genotype and presence of
nodules to allow inclusion of these 2 studies. Therefore,
our final meta-analysis consisted of 24 studies (22 with
IPD) and 3,272 patients (8,24–46).
Characteristics of the included studies are presented in Table 1. By definition, all studies evaluated
non-Hispanic Caucasian patients, the majority of
whom lived in European countries. The study designs
ranged from prospective cohort studies to a case series,
although one-half were of cross-sectional design and
one-third were case–control studies. IPD were available
for 92% of the studies. The mean duration of disease was
11 years, although 4 studies evaluated patients with early
disease (i.e., disease duration of ⬍2 years). Seventy-four
percent of the patients were female, and 76% of all patients
were seropositive for rheumatoid factor (RF).
A description of the criteria used for assessing
rheumatoid nodules was available for 14 (58%) of the
24 studies. Of these, 13 studies required physician
assessment of nodules by either direct clinical examination or chart review. A single study designated use of
patient self-report to identify RA patients with nodular
disease.
NO RELATIONSHIP OF RHEUMATOID NODULES TO THE HLA–DRB1 SE
757
Figure 1. Summary meta-analysis of association of the shared epitope (SE) with rheumatoid nodules. Comparison of 2 or 1 SE alleles versus 0 SE alleles (referent). The odds ratios are graphed on a log scale. The relative
weight of each study in the meta-analysis is indicated by the size of the associated box. The studies are listed
according to ethnic subgroup (as in Table 1). 95% CI ⫽ 95% confidence interval. † ⫽ P ⫽ 0.08 for heterogeneity.
A comparison of patients with nodules and
those without nodules (Table 2) revealed that patients
with nodular RA were significantly more likely to be
male, have a younger age at disease onset, and a longer
disease duration compared with patients without nodules (P ⬍ 0.001 for each). Additionally, patients with
nodular RA were also significantly more likely to be
seropositive for RF and to exhibit erosive disease (P ⬍
0.001 for each).
Relationship of rheumatoid nodules to presence
and dose of the SE. As shown in Figure 1, evaluation of the
presence of the SE (i.e., 1 or 2 versus 0 alleles) did not
demonstrate a significant association with disease (OR 1.3,
95% CI 0.97–1.6), although significant heterogeneity was
present (P ⫽ 0.08). Because 2 studies were observed to
have noticeably divergent results, these were excluded in a
sensitivity analysis. After this exclusion the effect estimate
remained stable (OR 1.3, 95% CI 1.1–1.6), although heterogeneity was reduced (P ⫽ 0.2). Of note, the divergent
results did not appear to be explained by a particular study
design, ethnic subgroup, or clinical characteristics such as
sex, RF positivity, or erosive disease.
Additional analyses of the association between
the SE dose and the presence of nodules failed to
provide evidence for a dose-dependent relationship,
because the estimate of the association for either 1 or 2
SE alleles was comparable (Table 3). Additionally, there
was no significant difference when inheritance of 2 SE
alleles was directly compared with that of a single allele
(OR 1.1, 95% CI 0.9–1.3).
To assess the influence of study quality, we
repeated the analysis excluding the cohort studies. This
758
GORMAN ET AL
Table 3. Association of HLA–DRB1 SE genotype with rheumatoid nodules*
Genotype
No. of
studies
No. of
patients with
genotype
Total no. of patients
available for analysis
OR (95% CI)
P for
heterogeneity†
2 SE alleles‡
401/401
404/404
401/404
101/401
101/404
101/101
1 SE allele
401/X
404/X
405/X
408/X
101/X
1001/X
23
18
8
16
16
11
6
24
20
18
17
14
16
17
801
130
17
114
145
40
17
1,604
647
167
62
33
376
50
1,606
826
454
749
786
519
391
2,414
1,384
860
739
625
1,017
703
1.3 (0.9–1.8)
1.1 (0.7–1.6)
1.5 (0.6–3.5)
1.3 (0.8–2.0)
1.2 (0.8–1.8)
0.9 (0.5–1.8)
0.9 (0.3–2.6)
1.2 (1.0–1.5)§
1.4 (1.1–1.8)¶
1.3 (0.9–1.9)
1.1 (0.6–2.0)
1.9 (0.9–3.8)
1.0 (0.8–1.4)
1.7 (0.96–3.1)
0.06
0.69
0.31
0.74
0.13
0.90
0.89
0.39
0.36
0.10
0.80
0.98
0.28
0.80
* The referent group for all comparisons was 0 shared epitope (SE) alleles. OR ⫽ odds ratio; 95% CI ⫽ 95% confidence
interval; X ⫽ non–SE allele.
† Values ⬍0.1 represent significant heterogeneity.
‡ All applicable genotypes combined.
§ P ⬍ 0.04.
¶ P ⬍ 0.008.
approach would be expected to result in an inflation of
the effect estimate, because cross-sectional and case–
control studies tend to demonstrate greater ORs compared with cohort studies; however, the estimate generated (OR 1.3, 95% CI 1.1–1.6) was strikingly similar to
that for our primary analysis.
Because of the prominent association of the SE
with nodules that is observed in patients with familial
RA (28,31,47,48), an analysis excluding 3 studies that
enrolled exclusively multiplex RA cases was performed
and revealed results similar to those from our main
analysis (OR 1.2, 95% CI 1.1–1.6). Moreover, the results
remained unchanged when stratified analyses for males
(OR 1.3, 95% CI 0.8–1.9) and cases of seronegative RA
(OR 1.0, 95% CI 0.7–1.6) were performed.
Relationship of rheumatoid nodules to the DRB1
SE genotype. Because individual SE alleles and genotypes may differ in their association with specific outcomes (49), an analysis by individual genotypes was
undertaken (Table 3). A total of 2,863 patients were
available for analysis by SE genotype. Only one genotype demonstrated a statistically significant association
with nodules, specifically that of a single DRB1*0401 SE
allele (OR 1.4, 95% CI 1.1–1.8). The effect estimates for
the other genotypes were relatively modest (i.e., OR ⬍
2.0), and those with greater estimated ORs tended to
have much broader confidence intervals (i.e.,
DRB1*0404 homozygotes, or a single DRB1*0408 or
*1001 SE allele). Because of a prior analysis in which we
demonstrated heterogeneity in the association of the SE
with erosions among Greeks in comparison with other
Caucasians (50), we performed a sensitivity analysis
excluding individuals of Mediterranean descent. There
was only one notable difference in this sensitivity analysis: specifically, the demonstration of an association of
rheumatoid nodules with inheritance of a single
DRB1*1001 allele (OR 2.4, 95% CI 1.3–4.7). In order to
evaluate a more ethnically homogenous population, a
separate analysis of HLA–DRB1 SE genotypes limited
to northern European Caucasians, the largest ethnic
subgroup (with 1,270 patients), was performed and
yielded results similar to those of the analyses of all
Caucasians combined.
In order to assess whether the inclusion of studies
comprising patients with early RA may have negatively
influenced the association of rheumatoid nodules with
genotypes, 2 separate analyses were performed. First,
meta-analyses with adjustment for disease duration were
performed for studies that provided IPD. These analyses
demonstrated results similar to those of the main analysis and resulted in more heterogeneity and less precise
results. In addition, a meta-analysis that excluded the 4
studies of patients with early RA was performed, with
results that corresponded to those of the main analysis.
Cumulative meta-analysis. To investigate the
strength of the evidence for an association of the SE with
nodules as it was collected over time, a cumulative
meta-analysis by year of publication was performed for
the genotype containing a single DRB1*0401 SE allele
(Figure 2). Results using this meta-analytic technique
NO RELATIONSHIP OF RHEUMATOID NODULES TO THE HLA–DRB1 SE
759
Figure 2. Cumulative meta-analysis of association of a single HLA–DRB1*0401 shared epitope allele with
rheumatoid nodules. The referent group was 0 shared epitope alleles. Studies are listed in order of the
publication date (from oldest to most recent). The odds ratios are graphed on a log scale.
revealed the most striking association with nodules
initially, with regression to the eventual modest estimate
(i.e., OR 1.4) upon the addition of relatively few studies.
Sensitivity analysis by definition of rheumatoid
nodules. In order to determine whether different definitions for rheumatoid nodules used by individual studies influenced our results, we performed a sensitivity
analysis excluding studies that either did not describe
how rheumatoid nodules were defined or that evaluated
the presence of nodules by patient report. There was no
considerable change in the results, although 3 SE genotypes with borderline significance in the unadjusted
analyses did achieve statistical significance in this sensitivity analysis. These genotypes included the presence of
the SE (OR 1.4, 95% CI 1.1–1.8), the presence of 2 SE
alleles (OR 1.5, 95% CI 1.1–2.1), and the presence of
DRB1*1001/X (OR 2.4, 95% CI 1.1–5.3).
Assessment of publication bias. A funnel plot of
the association of SE positivity with rheumatoid nodules
appeared relatively symmetric (data not shown). Moreover, statistical assessment by the regression asymmetry
test described by Egger et al (21) did not reveal significant evidence for publication bias (P ⫽ 0.8).
DISCUSSION
Results from this meta-analysis suggest that, at
least among Caucasians patients, no substantial associ-
ation exists between the SE and rheumatoid nodules. Of
all the genotypes examined, only those with a single
DRB1*0401 SE allele demonstrated a significant association with nodules, and this relationship was extremely
modest. Moreover, there was no evidence to suggest a
significant association of nodules with a double dose of
the SE or a contribution of DRB1*0401 homozygotes or
*0401/*0404 compound heterozygotes above that of
other genotypes, as has been suggested by the results of
smaller individual studies (28).
The results from our meta-analysis are intriguing,
because it has been widely regarded that the most
striking associations of the SE with RA severity exist
with rheumatoid nodules and other extraarticular manifestations (51). However, although several studies of
Caucasian patients support such a relationship (5,31,52),
many investigations have failed to replicate these findings (53–55). Moreover, it should not be surprising that
the genetic relationship with rheumatoid nodules may be
more complex than one that involves only DRB1 alleles.
For example, genetic interaction has recently been demonstrated between a tumor necrosis factor microsatellite
allele and DRB1*0401 homozygosity in nodular RA
(56). Additionally, linkage disequilibrium, the nonrandom association of alleles at different loci, is particularly
striking across the HLA region and must be considered
when interpreting genetic associations with disease.
760
Furthermore, the importance of environmental
factors and gene–environmental interactions in the development of rheumatoid nodules should not be ignored
and is beginning to be delineated. Specifically, Mattey et
al (57) recently described the relationship of a drugmetabolizing enzyme and cigarette smoke to RA severity. In that study, the presence of the null GSTM1
genotype was significantly associated with RF positivity
and titer and more severe erosive disease in RA patients
who smoked cigarettes. This finding is of particular
interest given the well-described association of cigarette
smoking with the development of rheumatoid nodules
(56,58,59). Another environmental, or specifically xenobiotic, risk factor known to be associated with the
development of rheumatoid nodules is methotrexate. It
is estimated that drug-induced nodules will develop in
⬃10% of methotrexate-treated patients with RA (60);
such nodules are histopathologically similar to spontaneously occurring nodules. Interestingly, methotrexateinduced nodules have been associated with the
DRB1*0401 allele, although in the same study a genetic
association of this allele with spontaneous rheumatoid
nodules was not observed (61).
There are several potential reasons why this
meta-analysis did not confirm an association of nodules
with the SE or particular DRB1 alleles. A common
problem in all clinical research relates to false-positive
results (Type I errors). In fact, this point has been
elegantly illustrated by a cumulative meta-analysis of
genetic association studies, in which initial studies
tended to demonstrate more striking results than did
subsequent investigations, particularly when the earliest
studies included fewer than 150 individuals and more
than 15 subsequent publications followed (62). Results
from the extensive cumulative meta-analyses performed
by Ioannidis et al (62), supported by our analyses
performed for DRB1*0401 and nodules, reinforce the
notion that caution should be exercised when interpreting results of individual genetic association studies, and
that meta-analysis can serve as a useful tool in the
systematic evaluation of any purported relationship.
An additional possibility is that we had insufficient power to detect a significant association of the SE
genotype with rheumatoid nodules. In fact, the only
DRB1 SE genotype to demonstrate statistical significance was also the largest (647 patients with a single
*0401 allele [1,384 patients total]). Unfortunately, in
contrast to methods available for determining the required sample size in other study designs, no widely
accepted approach for estimating the power of a metaanalysis exists, and, in fact, the primary concern regarding meta-analyses is overrepresentation of significant
GORMAN ET AL
results due to publication bias (63). Of interest, from our
findings we would estimate that an individual study
would require in excess of 400 cases and 400 controls to
demonstrate with 80% power the same effect size for the
association with a single DRB1*0401 allele.
Other potential reasons for our negative results
include variation within the population studied, such as
ethnic heterogeneity, differences in patient selection, or
imprecision in the definition of nodules. For example,
although our meta-analysis was performed among Caucasians, the most striking results were mainly limited to
a more defined subgroup, that of northern European
Caucasians (28,31). Moreover, studies examining individuals with early RA could lead to misclassification of
patients, because in cases categorized as non-nodular,
insufficient time may have elapsed in order for nodular
disease to develop. In addition, studies relying on patient
self-report of nodules may not be as sensitive and/or
specific compared with those incorporating physical
examination and/or medical record review. It is also
possible that systematic differences existed between
studies demonstrating positive versus negative results.
Because of these concerns, we performed several separate analyses to examine these individual possibilities.
This included subgroup analyses limited to northern
Europeans and those that used more precise definitions
of rheumatoid nodules, adjusted analyses incorporating
information regarding disease duration, and analyses
that excluded RA patients with familial disease. In all
cases, because there were no considerable differences
between these results and those from the primary analysis, we do not believe that our negative findings were
greatly influenced by these issues.
In summary, this IPD meta-analysis represents a
powerful and systematic approach to investigating the
association of the SE with rheumatoid nodules. Because
of the generous contribution of IPD regarding 2,925 RA
patients from 10 different countries by 22 researchers,
we were able to perform more sophisticated and detailed
analyses than is possible using traditional meta-analytic
methods based on published summary data. This approach allowed us to fully investigate associations with
the SE and the SE allelic genotype, and our results do
not support a significant association of rheumatoid
nodules with either. A more complete understanding of
the pathogenetic mechanisms involved in the development of nodules will likely require consideration of
other genetic and nongenetic influences. However, due
to the relative infrequency of rheumatoid nodules, individual genotypes and particular exposures, collaborative
and/or prospective pooled-analytic methods will be essential in further characterizing these associations.
NO RELATIONSHIP OF RHEUMATOID NODULES TO THE HLA–DRB1 SE
ACKNOWLEDGMENTS
We thank Kirsten A. Pfeiffer, BA, Glenys Thomson,
PhD, John J. Chen, PhD, Maria E. Suarez-Almazor, MD, PhD,
Catherine Mallon, RN, John M. Colford, MD, MPH, PhD, and
Mark Segal, PhD, for their insight and assistance with this
project. We are indebted to numerous colleagues who generously contributed data from their independent research for
inclusion in this meta-analysis. These contributors include
Graciela S. Alarcón, MD, MPH, Birmingham, AL; Sylvette
Bas, PhD, Geneva, Switzerland; Alain Cantagrel, MD, Toulouse, France; Bernard Combe, MD, PhD, Montpellier,
France; Bernard Cortet, MD, PhD, Lille, France; Anne Crilly,
PhD, Glasgow, Scotland; Anne Davidson, MB, BS, FRACP,
Bronx, NY; Kerstin Eberhardt, MD, PhD, Lund, Sweden;
Jean-François Eliaou, MD, PhD, Montpellier, France; Markku
Hakala, MD, Heinola, Finland; Ho-Youn Kim, MD, Seoul,
Korea; Alex J. MacGregor, MD, London, UK; Loreto Massardo, MD, Santiago, Chile; Janet E. McDonagh, MD, MRCP,
Birmingham, UK; Lachy McLean, MD, PhD, Loughborough,
UK; Fiona M. McQueen, MBChB, MD, FRACP, Auckland,
New Zealand; Olivier Meyer, MD, Paris, France; Timo Möttönen, MD, Paimio, Finland; Haralampos M. Moutsopoulos,
MD, FACP, FRCP (Edin), Athens, Greece; George Moxley,
MD, Richmond, VA; Antonio Núñez-Roldan, MD, PhD,
Seville, Spain; James R. O’Dell, MD, Omaha, NB; William
E. R. Ollier, PhD, Manchester, UK; Aleth Perdriger, MD,
PhD, Rennes, France; Janet E. Pope, MD, MPH, FRCPC,
London, Ontario, Canada; Jean Roudier, MD, PhD, Marseilles, France; Merrill Rowley, PhD, Victoria, Australia; Güher Saruhan-Direskeneli, MD, Istanbul, Turkey; Michael F.
Seldin, MD, PhD, Davis, CA; Alan J. Silman, MD, Manchester, UK; Dharam P. Singal, PhD, Hamilton, Ontario, Canada;
Fujio Takeuchi, MD, Tokyo, Japan; Wendy Thomson, PhD,
Manchester, UK; Pierre Tiberghien, MD, PhD, Besançon,
France; Yoshitaka Toda, MD, Osaka, Japan; Éric Toussirot,
MD, Besançon, France; Carl G. Turesson, MD, PhD, Malmö,
Sweden; Cor L. Verweij, PhD, Amsterdam, The Netherlands;
Daniel Wendling, MD, PhD, Besançon, France; Jehsye-Hsien
Yen, MD, Kaohsiung City, Taiwan; Steven A. Young-Min, BA,
BMBCh, MRCP, Newcastle-upon-Tyne, UK.
REFERENCES
1. Vlak T. Incidence of rheumatoid nodule in Dalmatia: similarities
and differences among populations. Arch Med Res 2003;34:56–9.
2. Turesson C, O’Fallon WM, Crowson CS, Gabriel SE, Matteson
EL. Occurrence of extraarticular disease manifestations is associated with excess mortality in a community based cohort of patients
with rheumatoid arthritis. J Rheumatol 2002;29:62–7.
3. Gabriel SE, Crowson CS, Kremers HM, Doran MF, Turesson C,
O’Fallon WM, et al. Survival in rheumatoid arthritis: a populationbased analysis of trends over 40 years. Arthritis Rheum 2003;48:
54–8.
4. 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:
1205–13.
5. Weyand CM, Hicok KC, Conn DL, Goronzy JJ. The influence of
HLA-DRB1 genes on disease severity in rheumatoid arthritis. Ann
Intern Med 1992;117:801–6.
6. Hakala M, Silvennoinen-Kassinen S, Ikaheimo I, Isosomppi J,
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
761
Tiilikainen A. HLA markers in a community-based rheumatoid
arthritis series. Ann Med 1997;29:291–6.
Salvarani C, Macchioni PL, Mantovani W, Bragliani M, Collina E,
Cremonesi T, et al. HLA-DRB1 alleles associated with rheumatoid arthritis in Northern Italy: correlation with disease severity.
Br J Rheumatol 1998;37:165–9.
Turesson C, Jacobsson L, Bergstrom U, Truedsson L, Sturfelt G.
Predictors of extra-articular manifestations in rheumatoid arthritis. Scand J Rheumatol 2000;29:358–64.
Egger M, Smith GD, Altman DG. Systematic reviews in health
care: meta-analysis in context. 2nd ed. London: BMJ; 2001. p.
313–35.
Reveille JD. The genetic contribution to the pathogenesis of
rheumatoid arthritis. Curr Opin Rheumatol 1998;10:187–200.
Apple RJ, Erlich H. HLA class II genes: structure and diversity.
In: Browning MJ, McMichael AJ, editors. HLA and MHC: genes,
molecules and function. Oxford (UK): BIOS Scientific Publishers,
Ltd; 1996. p. 100.
Egger M, Juni P, Bartlett C, Holenstein F, Sterne J. How
important are comprehensive literature searches and the assessment of trial quality in systematic reviews? Empirical study. Health
Technol Assess 2003;7:1–76.
Juni P, Witschi A, Bloch R, Egger M. The hazards of scoring the
quality of clinical trials for meta-analysis. JAMA 1999;282:
1054–60.
Conn VS, Rantz MJ. Research methods: managing primary study
quality in meta-analyses. Res Nurs Health 2003;26:322–33.
Hill CL, LaValley MP, Felson DT. Discrepancy between published
report and actual conduct of randomized clinical trials. J Clin
Epidemiol 2002;55:783–6.
Huwiler-Muntener K, Juni P, Junker C, Egger M. Quality of
reporting of randomized trials as a measure of methodologic
quality. JAMA 2002;287:2801–4.
Mantel N, Haenszel W. Statistical aspects of the analysis of data
from retrospective studies of disease. J Natl Cancer Inst 1959;22:
719–48.
Breslow NE, Day NE. Statistical methods in cancer research.
Volume I: the analysis of case-control studies. IARC Sci Publ
1980;32:5–338.
Rothman KJ, Greenland S. Approaches to statistical analysis. In:
Rothman KJ, Greenland S, editors. Modern epidemiology. 2nd ed.
Philadelphia: Lippincott-Raven; 1998. p. 190–1.
Sterne JA, Egger M. Funnel plots for detecting bias in metaanalysis: guidelines on choice of axis. J Clin Epidemiol 2001;54:
1046–55.
Egger M, Davey Smith G, Schneider M, Minder C. Bias in
meta-analysis detected by a simple, graphical test. BMJ 1997;315:
629–34.
Sterne JA, Gavaghan D, Egger M. Publication and related bias in
meta-analysis: power of statistical tests and prevalence in the
literature. J Clin Epidemiol 2000;53:1119–29.
Stata statistical software: release 8.0. College Station (TX): Stata
Corporation, 2003.
Eberhardt K, Fex E. Clinical course and remission rate in patients
with early rheumatoid arthritis: relationship to outcome after 5
years. Br J Rheumatol 1998;37:1324–9.
Vaidya B, Pearce SH, Charlton S, Marshall N, Rowan AD,
Griffiths ID, et al. An association between the CTLA4 exon 1
polymorphism and early rheumatoid arthritis with autoimmune
endocrinopathies. Rheumatology (Oxford) 2002;41:180–3.
Thomson W, Harrison B, Ollier B, Wiles N, Payton T, Barrett J, et
al. Quantifying the exact role of HLA–DRB1 alleles in susceptibility to inflammatory polyarthritis: results from a large, population-based study. Arthritis Rheum 1999;42:757–62.
Cortet B, Perez T, Roux N, Flipo RM, Duquesnoy B, Delcambre
B, et al. Pulmonary function tests and high resolution computed
tomography of the lungs in patients with rheumatoid arthritis. Ann
Rheum Dis 1997;56:596–600.
McDonagh JE, Dunn A, Ollier WE, Walker DJ. Compound
762
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
44.
45.
GORMAN ET AL
heterozygosity for the shared epitope and the risk and severity of
rheumatoid arthritis in extended pedigrees. Br J Rheumatol
1997;36:322–7.
Perdriger A, Chales G, Semana G, Guggenbuhl P, Meyer O,
Quillivic F, et al. Role of HLA-DR-DR and DR-DQ associations
in the expression of extraarticular manifestations and rheumatoid
factor in rheumatoid arthritis. J Rheumatol 1997;24:1272–6.
Hall FC, Weeks DE, Camilleri JP, Williams LA, Amos N, Darke
C, et al. Influence of the HLA-DRB1 locus on susceptibility and
severity in rheumatoid arthritis. QJM 1996;89:821–9.
MacGregor A, Ollier W, Thomson W, Jawaheer D, Silman A.
HLA-DRB1*0401/0404 genotype and rheumatoid arthritis: increased association in men, young age at onset, and disease
severity. J Rheumatol 1995;22:1032–6.
Bas S, Perneger TV, Mikhnevitch E, Seitz M, Tiercy JM, RouxLombard P, et al. Association of rheumatoid factors and antifilaggrin antibodies with severity of erosions in rheumatoid arthritis. Rheumatology (Oxford) 2000;39:1082–8.
Seidl C, Kasser UR, Fischer B, Koch U, Meier L, Fischer P, et al.
HLA-DR/DQ interaction in patients with erosive rheumatoid
arthritis presenting articular and extraarticular disease manifestations. Eur J Immunogenet 1999;26:19–27.
Hayem G, Chazerain P, Combe B, Elias A, Haim T, Nicaise P, et
al. Anti-Sa antibody is an accurate diagnostic and prognostic
marker in adult rheumatoid arthritis [published erratum appears
in J Rheumatol 1999;26:2069). J Rheumatol 1999;26:7–13.
Toussirot E, Auge B, Tiberghien P, Chabod J, Cedoz JP, Wendling
D. HLA-DRB1 alleles and shared amino acid sequences in disease
susceptibility and severity in patients from eastern France with
rheumatoid arthritis. J Rheumatol 1999;26:1446–51.
Cantagrel A, Navaux F, Loubet-Lescoulié P, Nourhashemi F,
Enault G, Abbal M, et al. Interleukin-1␤, interleukin-1 receptor
antagonist, interleukin-4, and interleukin-10 gene polymorphisms:
relationship to occurrence and severity of rheumatoid arthritis.
Arthritis Rheum 1999;42:1093–1100.
Benazet JF, Reviron D, Mercier P, Roux H, Roudier J. HLADRB1 alleles associated with rheumatoid arthritis in southern
France: absence of extraarticular disease despite expression of the
shared epitope. J Rheumatol 1995;22:607–10.
Saruhan-Direskeneli G, Istanbul Rheumatology Study Group.
Shared epitope homozygosity is strongly associated with rheumatoid arthritis in Turkey. Br J Rheumatol 1998;37:1126–8.
Stavropoulos C, Spyropoulou M, Koumantaki Y, Kappou I, Kaklamani V, Linos A, et al. HLA-DRB1 alleles in Greek rheumatoid
arthritis patients and their association with clinical characteristics.
Eur J Immunogenet 1997;24:265–74.
Boki KA, Drosos AA, Tzioufas AG, Lanchbury JS, Panayi GS,
Moutsopoulos HM. Examination of HLA-DR4 as a severity
marker for rheumatoid arthritis in Greek patients. Ann Rheum
Dis 1993;52:517–9.
Bali D, Gourley S, Kostyu DD, Goel N, Bruce I, Bell A, et al.
Genetic analysis of multiplex rheumatoid arthritis families. Genes
Immun 1999;1:28–36.
Meyer JM, Evans TI, Small RE, Redford TW, Han J, Singh R, et
al. HLA-DRB1 genotype influences risk for and severity of
rheumatoid arthritis. J Rheumatol 1999;26:1024–34.
Criswell LA, Mu H, Such CL, King MC. Inheritance of the shared
epitope and long-term outcomes of rheumatoid arthritis among
community-based Caucasian females. Genet Epidemiol 1998;15:
61–72.
Pope JE, Stevens A, Howson W, Bell DA. The development of
rheumatoid arthritis after recombinant hepatitis B vaccination.
J Rheumatol 1998;25:1687–93.
Rowley MJ, Stockman A, Brand CA, Tait BD, Rowley GL,
Sherritt MA, et al. The effect of HLA-DRB1 disease susceptibility
markers on the expression of RA. Scand J Rheumatol 1997;26:
448–55.
46. Combe B, Eliaou JF, Daures JP, Meyer O, Clot J, Sany J.
Prognostic factors in rheumatoid arthritis: comparative study of
two subsets of patients according to severity of articular damage.
Br J Rheumatol 1995;34:529–34.
47. MacGregor AJ, Snieder H, Rigby AS, Koskenvuo M, Kaprio J,
Aho K, et al. Characterizing the quantitative genetic contribution
to rheumatoid arthritis using data from twins. Arthritis Rheum
2000;43:30–7.
48. Jawaheer D, Lum RF, Amos CI, Gregersen PK, Criswell LA.
Influence of sex and shared epitope status on the severity of
familial RA [abstract]. Arthritis Rheum 2002;46 Suppl 9:S99.
49. Gorman JD, Criswell LA. The shared epitope and severity of
rheumatoid arthritis. Rheum Dis Clin North Am 2002;28:59–78.
50. Gorman JD, Chen JJ, Thomson G, Suarez-Almazor ME, Criswell
LA. Rheumatoid arthritis severity and dosage of the shared
epitope: results of a meta-analysis [abstract]. Arthritis Rheum
1999;42 Suppl 9:S85.
51. Weyand CM, McCarthy TG, Goronzy JJ. Correlation between
disease phenotype and genetic heterogeneity in rheumatoid arthritis. J Clin Invest 1995;95:2120–6.
52. Salvarani C, Macchioni P, Mantovani W, Rossi F, Veneziani M,
Boiardi L, et al. Extraarticular manifestations of rheumatoid
arthritis and HLA antigens in northern Italy. J Rheumatol 1992;
19:242–6.
53. Möttönen T, Paimela L, Leirisalo-Repo M, Kautiainen H, Ilonen
J, Hannonen P. Only high disease activity and positive rheumatoid
factor indicate poor prognosis in patients with early rheumatoid
arthritis treated with “sawtooth” strategy. Ann Rheum Dis 1998;
57:533–9.
54. Suarez-Almazor ME, Tao S, Moustarah F, Russell AS, Maksymowych W. HLA-DR1, DR4, and DRB1 disease related subtypes
in rheumatoid arthritis: association with susceptibility but not
severity in a city wide community based study. J Rheumatol
1995;22:2027–33.
55. Ioannidis JP, Tarassi K, Papadopoulos IA, Voulgari PV, Boki KA,
Papasteriades CA, et al. Shared epitopes and rheumatoid arthritis:
disease associations in Greece and meta-analysis of Mediterranean
European populations. Semin Arthritis Rheum 2002;31:361–70.
56. Mattey DL, Dawes PT, Fisher J, Brownfield A, Thomson W,
Hajeer AH, et al. Nodular disease in rheumatoid arthritis: association with cigarette smoking and HLA-DRB1/TNF gene interaction. J Rheumatol 2002;29:2313–8.
57. Mattey DL, Hutchinson D, Dawes PT, Nixon NB, Clarke S, Fisher
J, et al. Smoking and disease severity in rheumatoid arthritis:
association with polymorphism at the glutathione S-transferase
M1 locus. Arthritis Rheum 2002;46:640–6.
58. Wolfe F. The effect of smoking on clinical, laboratory, and
radiographic status in rheumatoid arthritis. J Rheumatol 2000;27:
630–7.
59. Harrison BJ, Silman AJ, Wiles NJ, Scott DG, Symmons DP. The
association of cigarette smoking with disease outcome in patients
with early inflammatory polyarthritis. Arthritis Rheum 2001;44:
323–30.
60. Combe B, Didry C, Gutierrez M, Anaya JM, Sany J. Accelerated
nodulosis and systemic manifestations during methotrexate therapy for rheumatoid arthritis. Eur J Med 1993;2:153–6.
61. Ahmed SS, Arnett FC, Smith CA, Ahn C, Reveille JD. The
HLA-DRB1*0401 allele and the development of methotrexateinduced accelerated rheumatoid nodulosis: a follow-up study of 79
Caucasian patients with rheumatoid arthritis. Medicine (Baltimore) 2001;80:271–8.
62. Ioannidis JP, Ntzani EE, Trikalinos TA, Contopoulos-Ioannidis
DG. Replication validity of genetic association studies. Nat Genet
2001;29:306–9.
63. Sutton AJ, Duval SJ, Tweedie RL, Abrams KR, Jones DR.
Empirical assessment of effect of publication bias on meta-analyses. BMJ 2000;320:1574–7.
Документ
Категория
Без категории
Просмотров
1
Размер файла
106 Кб
Теги
data, meta, patients, epitopes, hladrb1, shared, associations, lack, arthritis, caucasia, nodulesan, analysis, rheumatoid, 3272, individual
1/--страниц
Пожаловаться на содержимое документа