To assess the impact of the FcγRIIA-R/H131 polymorphism on the risk for antiphospholipid syndrome (APS), both primary and secondary to systemic lupus erythematosus (SLE).
To assess the impact of the FcγRIIA-R/H131 polymorphism on the risk for antiphospholipid syndrome (APS), both primary and secondary to systemic lupus erythematosus (SLE).
This international meta-analysis combined data from 9 research teams. FcγRIIA-R/H131 genotypes were determined in 481 APS cases (206 with primary APS), 1,420 SLE controls, and 1,655 disease-free controls. Data were combined using fixed-effects and random-effects models.
Compared with disease-free controls, the RR genotype was enriched in the entire group of APS cases (odds ratio [OR] 1.65, 95% confidence interval [95% CI] 1.28–2.14); this was driven mostly by patients with secondary APS (OR 1.95, 95% CI 1.45–2.63). The excess of RR homozygotes but not heterozygotes among APS patients suggested a recessive mode of inheritance, rather than the additive model seen for SLE susceptibility, where RR conferred greatest risk, and RH intermediate risk, for SLE. This probably reflected the additional influence of another opposing genetic effect of HH homozygosity on APS predisposition (OR 0.72 for RH versus HH, 95% CI 0.55–0.96). Among SLE patients, those with APS were more frequently HH homozygotes than heterozygotes (OR 0.56 for RH versus HH, 95% CI 0.39–0.81). HH homozygosity also tended to predominate in primary APS compared with secondary APS (OR 0.50 for RR versus HH, 95% CI 0.25–0.99 by fixed-effects model). There was no significant between-study heterogeneity for any of these effects.
The FcγRIIA-R/H131 polymorphism is an important determinant of predisposition to APS, with different influences on SLE and APS susceptibility per se.
The presence of antibodies in blood (lupus anticoagulant or anticardiolipin antibodies [aCL]) that recognize phospholipids, phospholipid-binding proteins, or both has been associated with a thrombophilic disorder, the antiphospholipid syndrome (APS) (1, 2). This syndrome is characterized by recurrent vascular thromboses involving the venous, arterial, and placental circulation and may occur alone (primary APS) or in conjunction with other autoimmune disease (secondary APS) (1–3). Several mechanisms have been proposed for explaining the procoagulant state of APS (2). According to one hypothesis (4), antiphospholipid antibody (aPL) binding to protein–phospholipid complexes on platelets, endothelial cells, or other cells may result in their activation via crosslinking of Fcγ receptors of type IIa (FcγRIIa) (5, 6). This activation may induce a prothrombotic phenotype (4). If this hypothesis is true, FcγRIIA function may regulate APS pathogenesis.
FcγRIIa is widely expressed on hematopoietic cells, including neutrophils and mononuclear phagocytes. This isoform is the only FcγR expressed on platelets and endothelial cells (7). A common functional polymorphism of the FcγRIIA gene plays a particular role in the expression of IgG2-mediated antibody responses. The 2 allelic forms of FcγRIIA differ by a single amino acid at residue 131 (histidine or arginine). The H131 allele is essential for handling IgG2 immune complexes (8). This polymorphism has been proposed to be influential in a variety of autoimmune diseases (7). The low-binding R131 allele imparts a significant risk for systemic lupus erythematosus (SLE) (9). Binding to the H131 variant can initiate more pronounced monocyte, platelet, and endothelial cell activation in the context of an IgG2 immune response, inducing a prothrombotic phenotype. Since autoimmune disease–associated aCL show IgG2 predominance (10, 11), one might expect an enrichment of the H131 allele in APS patients (11). However, the data are not yet conclusive (12, 13). Besides, in patients with APS secondary to SLE, this anticipated selection of the H131 allele may not be evident, given the overrepresentation of the RR genotype in SLE (9). Thus, the role of the FcγRIIA-R/H131 polymorphism in APS is still unclear.
APS is relatively uncommon, and isolated studies regarding its genetic background are unlikely to be conclusive. Previous analyses for HLA-conferred APS susceptibility have required the combination of data from various study groups (14). The aim of the present study was to investigate the importance of FcγRIIA alleles for APS susceptibility in the context of an international collaborative meta-analysis. Such an approach enhances the power to detect modest, but clinically important, differences between groups and helps to avoid spurious findings due to inconsistencies of the data from different research teams.
The meta-analysis included study groups in which FcγRIIA-R/H131 genotypes had been determined by molecular methods. The study groups consisted of APS patients (primary APS, APS secondary to SLE, or both) as well as disease-free control subjects and/or patients with SLE but without APS (control patients).
Participating investigators complied with the following rules to ensure consistency. The groups being compared should be racially matched. For studies of mixed racial descent, data for subjects of European, African, and Asian descent should be separated. Disease-free controls should not have APS or SLE. SLE should be defined according to the American College of Rheumatology 1982 revised criteria (15). APS should preferably be defined according to the 1999 preliminary criteria (16), but earlier alternative classification criteria were acceptable if they were clearly prespecified. Patients were classified as having secondary APS if they fulfilled criteria for both SLE and APS (15, 16). Patients with primary APS should not fulfill classification criteria for SLE or any other autoimmune disease. Patients in the SLE control group should not meet criteria for definite APS (16).
Research teams worldwide working on the low-affinity FcγR and their association with autoimmune diseases were invited to contribute data, provided that their study patients met the eligibility criteria defined above. We contacted teams identified as part of a previous meta-analysis that examined the role of the FcγRIIA-R/H131 polymorphism in SLE and lupus nephritis (9). Other potentially relevant studies were sought by searches of the Medline and EMBase databases (last search August 2002), with various combinations of key words (“antiphospholipid,” “anticardiolipin,” “polymorphism,” “allele,” “genetics,” “Fc receptor,” and “Fcγ receptor”). Finally, this strategy was supplemented by extensive communication with field experts.
Research teams from 9 centers (5 European, 3 Asian, and 1 American) agreed to participate. FcγRIIA genotype analysis was performed using polymerase chain reaction (PCR)–allele-specific oligonucleotide hybridization (8, 13, 17, 18), PCR–restriction fragment length polymorphism (12, 19), sequence-specific primer–PCR (20, 21), or amplification-refractory mutation screening–PCR (22). Databases were assembled and assessed at the coordinating center (Department of Hygeine and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece). Queries were clarified through communications with the participating investigators.
The main analysis used 4 different comparisons: APS versus disease-free controls (with separate analyses for primary and secondary APS), secondary APS versus SLE controls, primary versus secondary APS, and SLE controls versus disease-free controls. For each comparison, the following genotype contrasts were evaluated: RR versus RH and HH combined; RR and RH combined versus HH; RR versus RH; RH versus HH; and RR versus HH. The first contrast corresponds to a net recessive genetic effect of the R131 allele, the second contrast corresponds to a net dominant effect of this allele, and the other 3 contrasts probe into dose-response relationships. Odds ratios (ORs) and 95% confidence intervals (95% CIs) were estimated from each comparison and genotype contrast, since the studies typically used case–control designs.
For each comparison and genetic contrast, we estimated between-study heterogeneity using the Q statistic, which was considered significant at P < 0.10 (23). Study-specific data were combined using both fixed-effects (Mantel-Haenszel) (24) and random-effects (DerSimonian and Laird) (25) methods. The latter incorporate between-study heterogeneity and provide wider confidence intervals when the results of combined studies differ among themselves. In the absence of between-study heterogeneity, the two methods provide similar results. Random effects are more appropriate when between-study heterogeneity is present, and they are used as the primary reported analysis, unless stated otherwise.
Sensitivity analyses were limited to those studies that used the preliminary classification criteria for APS (16). In a different sensitivity analysis, we also examined whether ORs would be different if the observed genotype frequencies in disease-free controls were adjusted to those expected under the assumption of Hardy-Weinberg equilibrium (26). Results were similar (data not shown).
Finally, we performed cumulative meta-analysis and recursive cumulative meta-analysis (27, 28) to evaluate whether the combined OR changed over time as more data were accumulated for each comparison and contrast. Inverted funnel plots (29) were examined as diagnostics for heterogeneity related to the sample size of each study.
Analyses were conducted with the use of SPSS software, version 10.0 (SPSS, Chicago, IL) and Meta-Analyst software (Joseph Lau, Tufts–New England Medical Center, Boston, MA). P values are 2-tailed.
The meta-analysis included FcγRIIA-R/H131 genotyping of 481 patients with APS (206 with primary APS, 275 with secondary APS), and 3,075 control subjects (1,420 with SLE and 1,655 disease-free). The primary APS patients were from 3 participating research teams, while all teams contributed secondary APS cases (Table 1). Most APS subjects were of European descent (419 of 481); 52 APS patients were of Asian descent and 10 of African descent. All but 3 patients with primary APS were of European descent.
|Research team||Ethnic descent of subjects||APS patients||Controls||H131 allele frequency|
|All||Primary||Secondary||SLE patients||Healthy subjects||All APS patients||Healthy controls|
Three of the research teams used criteria (11, 30, 31) other than the preliminary criteria for the classification of APS. Apart from the essential clinical features of vascular thrombosis and complications of pregnancy (16), the alternative criteria allowed patients with thrombocytopenia (in 3 studies) or transient ischemic attacks (in 1 study) to be classified as having APS if they also had aCL or lupus anticoagulant. The distribution of genotypes (Table 2) in all disease-free control groups was in accordance with Hardy-Weinberg equilibrium (all P > 0.10).
|Research team||Ethnic descent of subjects||FcγRIIA-R/R131||FcγRIIA-R/H131||FcγRIIA-H/H131|
|All APS patients||SLE controls||Healthy controls||All APS patients||SLE controls||Healthy controls||All APS patients||SLE controls||Healthy controls|
The APS patient group had an enrichment of the homozygous state for the low-binding allele, with an OR for APS being 1.65 in RR homozygous patients compared with both other genotypes combined (P < 0.001; no significant between-study heterogeneity) (Figure 1A). Evaluation of specific genotype contrasts showed that the genetic relationship was complex. The presence of RR increased APS risk as compared with RH, and less so as compared with HH, although the 95% CIs overlapped. The presence of RH significantly decreased the risk for APS as compared with HH (Figure 1B).
|Group, genotype comparison||No. of comparisons||No. of subjects||REM OR (95% CI)||FEM OR (95% CI)|
|All APS vs. disease-free controls|
|RR vs. RH + HH||10||2,134||1.65 (1.28–2.14)||1.63 (1.25–2.11)|
|RR + RH vs. HH||10||2,134||0.90 (0.70–1.16)||0.90 (0.70–1.16)|
|RR vs. RH||10||1,404||1.86 (1.39–2.47)||1.83 (1.37–2.43)|
|RH vs. HH||10||1,686||0.72 (0.55–0.96)||0.72 (0.55–0.95)|
|RR vs. HH||10||1,178||1.38 (1.01–1.88)||1.37 (1.00–1.87)|
|Primary APS vs. disease-free controls|
|RR vs. RH + HH||4||672||1.17 (0.76–1.79)||1.17 (0.76–1.78)|
|RR + RH vs. HH||4||672||0.87 (0.59–1.29)||0.87 (0.59–1.28)|
|RR vs. RH||4||480||1.24 (0.79–1.96)||1.26 (0.80–1.99)|
|RH vs. HH||4||524||0.81 (0.53–1.23)||0.81 (0.53–1.22)|
|RR vs. HH||4||340||1.01 (0.61–1.67)||1.01 (0.62–1.67)|
|Secondary APS vs. disease-free controls|
|RR vs. RH + HH||10||1,928||1.95 (1.45–2.63)||1.91 (1.42–2.57)|
|RR + RH vs. HH||10||1,928||0.93 (0.69–1.27)||0.95 (0.70–1.28)|
|RR vs. RH||10||1,259||2.23 (1.60–3.10)||2.18 (1.56–3.03)|
|RH vs. HH||10||1,528||0.70 (0.50–0.98)||0.69 (0.50–0.97)|
|RR vs. HH||10||1,069||1.61 (1.12–2.32)||1.61 (1.12–2.32)|
|Secondary APS vs. SLE controls|
|RR vs. RH + HH||11||1,698||1.62 (1.18–2.24)||1.60 (1.17–2.19)|
|RR + RH vs. HH||11||1,698||0.72 (0.52–0.99)||0.72 (0.52–0.99)|
|RR vs. RH||11||1,196||1.97 (1.37–2.87)||1.99 (1.41–2.81)|
|RH vs. HH||11||1,274||0.56 (0.39–0.81)||0.56 (0.39–0.80)|
|RR vs. HH||11||926||1.06 (0.71–1.58)||1.05 (0.71–1.55)|
|Primary vs. secondary APS|
|RR vs. RH + HH||4||302||0.59 (0.30–1.17)||0.60 (0.35–1.04)|
|RR + RH vs. HH||4||302||0.64 (0.35–1.18)||0.63 (0.35–1.15)|
|RR vs. RH||4||220||0.64 (0.35–1.16)||0.66 (0.36–1.18)|
|RH vs. HH||4||221||0.73 (0.38–1.40)||0.71 (0.37–1.38)|
|RR vs. HH||4||163||0.50 (0.25–1.01)||0.50 (0.25–0.99)|
|SLE controls vs. disease-free controls|
|RR vs. RH + HH||10||3,072||1.36 (1.02–1.82)†||1.34 (1.10–1.63)|
|RR + RH vs. HH||10||3,072||1.36 (1.02–1.81)‡||1.29 (1.09–1.53)|
|RR vs. RH||10||2,064||1.28 (0.92–1.78)†||1.25 (1.02–1.54)|
|RH vs. HH||10||2,442||1.28 (0.93–1.77)‡||1.21 (1.01–1.44)|
|RR vs. HH||10||1,638||1.63 (1.17–2.27)†||1.61 (1.27–2.05)|
The primary APS patients did not differ significantly from the disease-free controls in any genotype contrast, but modest associations could have been missed due to limited data. Conversely, as compared with disease-free controls, secondary APS patients had an overrepresentation of the RR genotype relative to both of the other genotypes combined (OR 1.95) (Figure 2) or relative to each of the other genotypes separately (OR 2.23 versus RH and OR 1.61 versus HH). Again, HH homozygous SLE patients were at greater risk of developing APS compared with RH heterozygotes.
Among lupus patients, both homozygous states contributed to APS susceptibility. In particular, RR homozygotes were more prone to developing APS. A maximum effect was seen when RR homozygotes were contrasted with RH heterozygotes (OR 1.97). Likewise, the HH genotype was associated with a trend toward an increased predisposition to APS, particularly when contrasted with RH heterozygosity (Figure 3).
The HH genotype tended to be more common in patients with primary APS than in those with secondary APS. There was formal statistical significance with fixed-effects modeling when HH was contrasted against RR in particular, and there was no significant between-study heterogeneity for any genotype contrast.
SLE controls had an enrichment of the R131 allele compared with disease-free controls. RR homozygotes were at greater risk of developing SLE compared with RH heterozygotes (OR 1.28). A maximum effect was seen when RR homozygotes were contrasted with HH homozygotes (OR 1.63), while the OR for RH versus HH was 1.28. Even though significant between-study heterogeneity was detected in these comparisons, the results are consistent with the previously described dose-response effect of the R131 allele in SLE susceptibility (9).
When we excluded studies that did not use the 1999 preliminary criteria for the classification of APS, the results were similar. The random effects OR estimate for the risk of developing APS was 1.74 in RR homozygous subjects compared with both of the other genotypes combined (95% CI 1.26–2.41; no significant between-study heterogeneity). There was still an enrichment of the RR genotype in secondary APS patients as compared with disease-free controls (OR 2.04 for RR versus other genotypes combined, 95% CI 1.42–2.95; no significant heterogeneity). Among SLE patients, the homozygous HH genotype also conferred a greater risk of developing APS than did RH heterozygosity (OR 0.57 for RH versus HH, 95% CI 0.35–0.93; no significant heterogeneity).
There was no evidence that the strength of the observed associations was different in early studies versus those published recently. In particular, the effect of the R131 allele seemed to become clearer over time for both the total APS group and the group with APS secondary to SLE, while the effect of the high-binding allele had been steady all along (data not shown). Similarly, inverted-funnel plots showed no evidence of bias differentiating the magnitude of the observed effects in the key comparisons between small and large studies (data not shown).
This international meta-analysis suggests a complex genetic background underlying the relationship between the FcγRIIA-R/H131 polymorphism and APS. A significant increase in RR homozygosity was documented in the whole group of APS patients. This selection was most striking in patients with APS secondary to SLE. The R131 allele seemed to confer risk for APS under a recessive model, whereas the effect of R131 on susceptibility to SLE has been found to have a dose-response character (9). This difference may be explained by the fact that among lupus patients, those who have APS also have an overrepresentation of homozygosity for the high-binding H131 allele. This may be even more prominent in primary APS, but data on primary APS comparisons were too limited to be definitive. As a result of these composite genetic influences, when the whole group of APS patients was contrasted against disease-free controls, there was a selection of the HH genotype as compared with the RH genotype. Thus, the meta-analysis results suggest that the observed genetic profile may be a composite of 2 different and opposing influences with regard to APS susceptibility.
Recent evidence suggests an effect of FcγR as potential initiators of thrombosis. This complication seems to be a consequence of platelet activation initiated when platelet FcγRIIa are crosslinked by antibodies (5, 32). Since the interaction of IgG2-containing antibodies with FcγRIIa is allotype-dependent (8), it has been hypothesized that the high-binding H131 allele would be overrepresented among subjects with antibody-mediated thrombosis (11, 33). The hypothesis has also been investigated in heparin-induced thrombocytopenia, another syndrome with similar immune-mediated thrombosis, but no consistent relationship between the H131 allele and this syndrome was shown (34).
The results of this meta-analysis may help to explain discrepancies among findings of previous studies of FcγRIIA-R/H131 in APS (11–13). HH homozygosity increases the risk of APS relative to RH heterozygosity. However, the effect of HH homozygosity for susceptibility to APS is overwhelmed by the larger effect of RR homozygosity for susceptibility to SLE in general (9), especially among patients with secondary APS. The R131 allele may confer risk for SLE through deficient handling of IgG2-containing immune complexes by the mononuclear phagocyte system, leading to their tissue deposition and to accelerated organ damage (7).
It is not clear whether RR homozygosity may also confer an increased risk of primary APS per se, aside from SLE, through some common pathophysiologic link to SLE. We should caution that the classification criteria for APS and SLE are functional criteria and may not fully correspond to the subgrouping of APS and SLE based on genetic predisposition. APS is a remarkably heterogeneous syndrome with different prognostic profiles (35).
Alternative hypotheses could explain an independent effect of RR homozygosity on the risk of APS. For example, apoptotic cells are a major source of autoantigens, and an impairment of their physiologic clearance may promote the development of autoimmunity. Anionic phospholipids redistribute from the inner leaflet to the outer leaflet of cell membranes during apoptosis (36). This systemic exposure could enable the binding of phospholipid-binding proteins such as β2-glycoprotein I (β2GPI) to apoptotic cell membranes (37) and may also trigger the production of aPL antibodies (38, 39). Phospholipid–β2GPI complexes on the surface of membrane blebs are recognized by aPL antibodies (37, 38, 40, 41), which leads to opsonization of apoptotic cells that are then phagocytosed by FcR-positive macrophages (40, 41). Considering that aPL (10, 11), especially those with reactivity to β2GPI (42, 43), show IgG2-dominant distribution, such antibodies would be predicted to be poor opsonins in RR homozygous subjects. Defective clearance of aPL-opsonized apoptotic particles by macrophages may lead to inflammatory removal pathways (44, 45), favoring an autoimmune, rather than an antiinflammatory, response to apoptotic cells. Thus, antigen processing and presentation by antigen-presenting cells (44, 45) provide an antigenic stimulus for specific T and B clones, leading to further aPL antibody production that may exert procoagulant effects (2). Moreover, persistently circulating apoptotic cells could express procoagulant properties, thus supporting thrombotic events (46).
That the RR genotype is mostly enriched among lupus patients with APS is also consistent with this explanation, since SLE is characterized by an increased rate of activation-induced cell death (47). Increased apoptotic load with the augmented exposure of anionic phospholipids may amplify the consequences of the defective handling of apoptotic cells in lupus patients who are homozygous for the low-binding allele. Remarkably, aPL antibodies occur more frequently and earlier in SLE patients with the RR genotype (20, 48). Furthermore, macrophages from SLE patients with sufficient expression of receptors implicated in phagocyte recognition of cells undergoing apoptosis (CD14 and CD36) exhibit defective engulfment of apoptotic cell material in vitro (49).
Patients with either primary or secondary APS have similar clinical profiles as far as thrombotic manifestations are concerned (3, 50). Nevertheless, arthritis, low C4 levels, and hematologic abnormalities such as hemolytic anemia, thrombocytopenia, leukopenia, and neutropenia are more common among patients with APS secondary to SLE (3, 50). The selection of the RR homozygous state that was demonstrated only in patients with secondary APS relative to disease-free controls could also be related to these differences, even though other factors may also play a role in the pathogenesis of these manifestations. It is noteworthy that such clinical features and serologic findings seem to be overrepresented in lupus patients with low-binding FcγR alleles (20, 21). Moreover, a critical role for FcγR has been demonstrated in models of collagen-induced arthritis as well as in models of experimental cytopenias (51, 52).
The relevance of the FcγRIIA-R/H131 polymorphism for APS susceptibility should be viewed at the population level. Although the summary OR estimates suggest only a moderate genetic effect, the importance of this effect may be considerable at the population level, given the high frequency of the RR genotype (∼25%) in populations of European descent (7). Thus, modest ORs translate to a clinically meaningful proportion of APS cases that could be attributed to RR homozygosity (at least 10% in populations of European descent). Empirical evidence suggests that for multigenetic diseases, the magnitude of the ORs is generally modest. Among 55 genetic associations examined in different disciplines, none had an OR exceeding 2.0, and only 13 showed a significant OR exceeding 1.5 (53).
Some limitations of this study should be discussed. First, the number of APS patients with specific clinical manifestations was too small to reliably assess the effect of the FcγRIIA-R/H131 polymorphism on the risk of vascular thromboses or other APS-related features. Multiple comparisons and small subgroups would make such inferences impractical, even with the sample size of the meta-analysis. Second, bias is possible in a meta-analysis. However, bias diagnostics did not suggest the presence of such problems in this study. Three study teams did not use the preliminary criteria for the classification of APS; nevertheless, similar results were obtained when data from these study teams were excluded. Last, the association of the FcγRIIA-R/H131 polymorphism with APS could be explained by the existence of linkage disequilibrium between this gene and other candidate genes on chromosome 1 that may also be more directly relevant for the risk of specific disease manifestations (54). This would require the investigation of extended haplotypes (55, 56) in the future. Genetic variants in different genes might also contribute to the pathogenesis of APS (57, 58). Recognition of specific disease-associated genetic factors may expand our understanding of disease pathogenesis and may also be useful for identifying subjects at increased risk of developing APS.
Other key investigators of the collaborating study teams involved in this project were as follows: Toru Fukazawa, MD, PhD (Juntendo University, Tokyo, Japan), Hiroshi Hashimoto, MD, PhD (Juntendo University, Tokyo, Japan), Elisabeth Tournier-Lasserve, MD (Faculté de Médecine Necker, Paris, France), Kwang-Taek Oh, MD (Hanyang University, Seoul, South Korea), Jan G. J. van de Winkel, PhD (Department of Immunology and Genmab, University Medical Center Utrecht, Utrecht, The Netherlands).