HLA-DRB1 Genotypes and the Risk of Developing Anti Citrullinated Protein Antibody (ACPA) Positive Rheumatoid Arthritis

Objective To provide a table indicating the risk for developing anti citrullinated protein antibody (ACPA) positive rheumatoid arthritis (RA) according to one’s HLA-DRB1 genotype. Methods We HLA-DRB1 genotyped 857 patients with ACPA positive RA and 2178 controls from South Eastern and Eastern France and calculated Odds Ratios (OR) for developing RA for 106 of 132 possible genotypes accounting for 97% of subjects. Results HLA-DRB1 genotypic ORs for developing ACPA positive RA range from 28 to 0.19. HLA-DRB1 genotypes with HLA-DRB1*04SE (HLA-DRB1*0404, HLA-DRB1*0405, HLA-DRB1*0408), HLA-DRB1*04∶01, HLA-DRB1*01 are usually associated with high risk for developing RA. The second HLA-DRB1 allele in genotype somewhat modulates shared epitope associated risk. We did not identify any absolutely protective allele. Neither the Reviron, nor the du Montcel models accurately explains our data which are compatible with the shared epitope hypothesis and suggest a dosage effect among shared epitope positive HLA-DRB1 alleles, double dose genotypes carrying higher ORs than single dose genotypes. Conclusion HLA-DRB1 genotypic risk for developing ACPA positive RA is influenced by both HLA-DRB1 alleles in genotype. We provide an HLA-DRB1 genotypic risk table for ACPA positive RA.


Introduction
Susceptibility for developing rheumatoid arthritis (RA) is associated with particular HLA-DRB1 alleles like HLA-DRB1*04, HLA-DRB1*01 and HLA-DRB1*10 [1]. A molecular basis for this association was provided by Gregersen and al. who showed that RA associated HLA-DRB1 alleles contain a conserved 5 amino acid stretch, the ''shared epitope'' (SE) in the third hypervariable region of their DRB1 chain [2]. This lead to a simple model in which shared epitope positive HLA-DRB1 alleles carried susceptibility for developing RA and shared epitope negative alleles were considered neutral.
Since 1987, numerous studies have confirmed the association of RA with shared epitope positive HLA-DRB1 alleles. However, it has been suggested that a few shared epitope negative HLA-DRB1 alleles protect against the development of RA, whereas most others are neutral [3][4][5].
However, the validity of these models has not been established. Here, we decided to calculate the risk for developing RA given by any individual HLA-DRB1 genotype, with no a priori model.
Complexity of the RA/HLA-DRB1 association further increased with the discovery that RA can be divided into two subtypes according to the presence in a patient's serum of autoantibodies directed at citrullin residues on different proteins [6]. Indeed, the sera of two thirds of patients with RA contain antibodies to citrullinated proteins. Citrullin is an amino acid generated by the posttranslational modification of arginyl residues by peptidyl arginine deiminases. These autoantibodies are called anti citrullinated protein antibodies (ACPA) [6]. Presence or absence of ACPAs define two subtypes of RA. ACPA positive RA is well defined and the 2010 ACR criteria for the diagnosis of RA include a positive ACPA test [7]. ACPA negative RA is much more heterogeneous [8]. The association of RA with shared epitope positive HLA-DRB1 alleles is stronger in ACPA positive RA than in ACPA negative RA [8,9].
Here, we studied a series of 857 ACPA positive RA patients and compared them with a series of 2178 controls.
Every patient and control, all from South Eastern France was HLA-DR typed for 20 different HLA-DRB1 alleles. Bayesian statistics were used to define susceptible and protective genotypes and to calculate accurate confidence intervals for the associated Odds Ratios. We calculated Odds Ratios (OR) for 102 of 136 possible HLA-DRB1 genotypes, for which the number of patients plus controls was at least 5. Thirty genotypes had ORs significantly higher than 1 and were considered high risk, 45 had ORs not significantly different from 1 and were considered neutral, 27 genotypes had ORs significantly lower than 1 and were considered low risk.

Patients and Controls
Ethical approval was obtained for this study from CPP Marseille II. All participants gave their informed consent. All participants signed informed consent according to the Declaration of Helsinki.
In this cohort study, we took blood samples from 857 ACPA positive RA patients and 2178 controls from Marseilles (South Eastern France) and Besançon (Eastern France). RA patients fulfilled the 2010 EULAR/ACR criteria [7] and tested positive for anti cyclic-citrullinated peptide antibodies (ACPA) using the anti CCP2 kit which is routinely used in our patients with arthritis. The control group included voluntary blood or bone marrow donors from the same area. Sex ratio and age were not significantly different between patients and controls. Seventy one percent of patients were positive for at least one shared epitope positive HLA-DRB1 allele (Table 1).

ACPA and Rheumatoid Factor Testing
Positivity for anti citrullinated peptide antibodies (ACPA) was used to define mainstream, classical RA [8]. ACPA were detected by anti CCP2 Enzyme-linked immuno sorbent assay (ELISA) (Immunoscan RA, Euro-Diagnostica, Arnhem, the Netherlands). This kit is currently the most commonly used in Southern France and positivity is defined by a cutoff value of 25 Units/ml at a dilution of 1/50. 91% of our patients selected for ACPA positivity by anti CCP2 testing tested positive for rheumatoid factor by ELISA using the Orgentec Kit (Mainz, Germany) with a 20 Units/ml cutoff at a 1/100 dilution.

HLA-DRB1 Genotyping
Low resolution HLA-DRB1 typing was carried out according to the manufacturer's specification for LABType SSO (One Lambda Inc, USA) and the retrieved output was analyzed by HLA Fusion v 1.2.1. software (One Lambda Inc, USA) for allele identification. HLA-DRB1*04-positive samples were subtyped by PCR sequence-specific primers (Olerup SSP HLA-DRB1*04, Genovision, Vienna, Austria).
We The rationale for this regrouping is that these three HLA-DRB1*04 alleles encode a common aminoacid sequence, QRRAA in their third hypervariable region (HV3). This HV3 sequence is different from that encoded by HLA-DRB1*0401, QKRAA. Therefore, HLA-DRB1*0401 is not included in this HLA-DRB1*04 allelic subgroup that we called HLA-DRB1*04SE.
After this regrouping, we ended up with 16 allelic groups defining 136 genotypes.

Statistical Analysis
The statistical method used in this study to estimate the Odds Ratio and its confidence interval for the different genotypes is based on the Bayesian theory. This method is more stringent that the one using the maximization of the likelihood function especially when there are less than 5 individuals in one of the two classes of genotypes (patient or control). This method was applied with the MatlabH Sofware. Data were simulated according to the posteriori density of Bayesian statistics in order to obtain the confidence intervals of the different odds ratios.

HLA-DRB1 Alleles Associated with RA
Seventy one percent of patients and 41% of controls were positive for the HLA-DRB1 shared epitope. Fifty one percent of patients and 37% of controls expressed one shared epitope positive allele. Twenty percent of patients and 4.5% of controls expressed two shared epitope positive HLA-DRB1 alleles.

RA Associated and Non RA Associated HLA-DRB1 Alleles
In this study of HLA-DRB1 alleles expressed in 857 patients with ACPA positive RA and 2178 normal controls from South Eastern France, we confirm the classical association of RA with shared epitope positive HLA-DRB1 alleles.
Our data confirm the shared epitope hypothesis and its already known ''dose effect''. Indeed, if one considers classical shared epitope positive alleles HLA-DRB1*04SE, *04:01, *10 and *01, they define 10 ''double dose'' genotypes. Only 9 of these 10 genotypes have been considered for OR calculation (the homozygous DRB1*10 subjects were less than 5). All 9 had ORs significantly higher than 1 (high risk genotypes) (Figure 2). Among 48 ''single dose'' shared epitope positive genotypes, 40 were considered for OR calculations, of which 17 (40%) were ''high risk'', 22 were neutral and only 1, DRB1 *13/DRB1 *01, was ''low risk''. Finally, among 78 shared epitope negative genotypes, 53 were considered for OR calculation, of which 25 (47%) were ''low risk'', 24 (45%) were ''neutral'' and 4 (8%)'' high risk'' ( Figure 2). In short, we find that every ''double dose'', and 40% of ''single dose'' shared epitope positive genotypes are high risk, against only 8% of shared epitope negative genotypes. If one considers ORs associated with pooled shared epitope positive genotypes (genotypes obtained by considering shared epitope positive alleles as one category), then 14 of 17 pooled shared epitope positive genotypes are high risk and 3 are neutral ( Figure 2). Altogether, these data confirm validity of the shared epitope as the determinant of HLA-DRB1 susceptibility for developing RA. However, classifications of alleles according to models are based on a priori assumptions of how alleles interact and determine risk (control of peptide binding…). Here, we made no such assumption and just calculated Odds ratios for as many genotypes as we could. We provide a table that gives an idea of the risk for developing ACPA positive RA given one's HLA-DR genotype. Under its current form, it gives the risk for 102 of 136 genotypes, accounting for 97% of our population (Figure 2). It may be used to help diagnosis of undifferentiated arthritis, especially before the emergence of ACPA response. It may also prove useful for genetic counseling in families affected by rheumatoid arthritis, indicating which individual to monitor for emergence of ACPA response and possibly early treatment or even prevention.
Finally, our genotypic risk table does not necessarily reflects the effect and interaction of HLA-DRB1 genes only, but may also involve non HLA-DRB1 genes on HLA-DRB1 haplotypes.
The Remaining Third: ACPA Negative RA This study did not approach the very controversial issue of ACPA negative RA and its HLA-DRB1 association(s). Indeed, ACPA negative RA is somewhat heterogeneous and careful analysis of patients' files often ends up with diagnosis other than RA. Still, one large scale study of HLA-DRB1 associations in ACPA negative RA patients suggests that half of them test positive for rheumatoid factors and show HLA-DRB1 association similar to ACPA positive RA [11]. Such patients will be the objects of future studies.
Other Polymorphisms in the HLA Region may be Relevant to Susceptibility for Developing ACPA Positive RA A recent very large-scale association study using 3000 SNPs in the MHC confirmed the importance of polymorphisms in the HLA-DRB1 chain, pinpointing three polymorphic residues, two of which are included in the shared epitope region. However, this study did not consider the genotypic effect, that is how BOTH HLA-DRB1 alleles interact to modulate risk for developing RA [12]. Here, we did not focus on the identification of new polymorphic residues in HLA-DRB1 alleles but on the interaction of individual alleles inside genotypes to influence RA susceptibility.