Ce travail s’est ici déroulé en deux étapes principales.
Il était tout d’abord question d’étudier la pertinence de la nouvelle classification des allèles HLA-DRB1 en termes d’association avec la sévérité de la PR. La classification réalisée par Tezenas du Montcel et al., présentée dans la partie « Patients et Méthodes » de ce manuscrit, avait été développée avec l’objectif de redéfinir le rôle joué par cette région génomique dans la physiopathologie de la PR. Il était donc aussi pertinent, une fois cette nouvelle classification définie, de la tester pour tenter d’étudier la prédiction de la sévérité de la maladie.
Ce travail a été présenté dans la publication suivante :
Gourraud PA, Boyer JF, Barnetche T, Abbal M, Cambon-Thomsen A, Cantagrel A, et al. A new classification of HLA-DRB1 alleles differentiates predisposing and protective alleles for rheumatoid arthritis structural severity. Arthritis Rheum 2006;54(2):593-9.
Ensuite, bénéficiant des données du XIIIième Workshop International
d’Histocompatibilité sur les patients atteints de PR, l’objectif était de réaliser une méta-analyse sur les différents risques de développer la maladie retrouvés dans les échantillons provenant d’origines diverses. Ceci avec l’objectif de tester la nouvelle classification des allèles HLA-DRB1 sur des populations non-caucasoïdes.
Ceci a donné lieu à l’écriture d’un article soumis et refusé par la revue Arthritis and Rheumatism. Cet article est en cours de correction pour re-soumission
ARTHRITIS & RHEUMATISM Vol. 54, No. 2, February 2006, pp 593–599 DOI 10.1002/art.21630
© 2006, American College of Rheumatology
A New Classification of HLA–DRB1 Alleles Differentiates
Predisposing and Protective Alleles for Rheumatoid Arthritis
Structural Severity
Pierre-Antoine Gourraud,1 Jean-Fre´de´ric Boyer,2 Thomas Barnetche,1 Michel Abbal,3
Anne Cambon-Thomsen,1Alain Cantagrel,2 and Arnaud Constantin4
Objective. A new classification of HLA–DRB1
alleles supporting the shared epitope hypothesis of rheumatoid arthritis (RA) susceptibility was recently introduced. We investigated the relevance of this classi- fication in terms of the structural severity of RA.
Methods. The study group comprised 144 patients
who were included in a prospective longitudinal cohort of French Caucasoid patients with early RA. Progres- sion of the total radiographic damage score (Sharp/van der Heijde method) was used to quantify the structural severity of RA after 4 years of followup. HLA–DRB1 typing and subtyping were performed by polymerase chain reaction, using a panel of sequence-specific oligo- nucleotide probes. HLA–DRB1 alleles were classified according to the above-mentioned new system. The association between the HLA–DRB1 allele groups (S1, S2, S3P, S3D, and X) and the structural severity of RA
was analyzed with nonparametric statistical tests.
Results. The presence of S2 alleles (HLA–
DRB1*0401 and HLA–DRB1*1303) was associated with
severe forms of RA (P ⴝ 0.004); a significant dose effect was observed (P ⴝ 0.01). The presence of S3D alleles
(HLA–DRB1*11001, HLA–DRB1*1104, HLA–DRB1*12, and HLA–DRB1*16) was associated with benign forms of RA (P < 0.0001), and a significant dose effect was observed (P < 0.01).
Conclusion. The studied classification of HLA–
DRB1 alleles is relevant in terms of RA outcomes. Compared with a previously described classification system, this system differentiates predisposing (S2) and protective (S3D) alleles for RA structural severity, which, respectively, correspond to KRRAA and DRRAA amino acid patterns at position 70–74 of the third hypervariable region of the HLA–DR chain.
Rheumatoid arthritis (RA) is a common, com- plex genetic disease, and, despite a significant genetic component, no gene other than HLA–DRB1 has been clearly demonstrated to be involved (1). The association between HLA–DR4 and RA was reported more than 25 years ago (2). HLA–DRB1 alleles encoding a highly conserved amino acid pattern known as the shared epitope (SE), which is characterized by the RAA pattern at position 72–74 of the third hypervariable region of
different HLA–DR chains, are associated with RA
susceptibility (3). The presence and dose effect of these alleles affect the course and outcome of RA, because HLA–DRB1 alleles encoding the SE are reported to be associated with the structural severity of RA (4).
In consideration of the SE hypothesis, Gao et al proposed to further differentiate HLA–DRB1 alleles encoding the RAA pattern, taking into account the amino acids at positions 71 and 86 (5). The classification suggested by Gao et al stressed the significant differ- ences between HLA–DRB1 alleles in terms of suscepti- bility to RA (5).
Supported by INSERM Unit 558, the Universite´ Paul Saba- tier Toulouse III, the Hoˆpitaux de Toulouse, the Ecole Normale Supe´rieure de Lyon, and the Centre National de la Recherche Scientifique.
1Pierre-Antoine Gourraud, PhD, Thomas Barnetche, PhD,
Anne Cambon-Thomsen, MD: INSERM U558, Toulouse, France;
2Jean-Fre´de´ric Boyer, MD, Alain Cantagrel, MD: Centre Hospitalier
Universitaire Rangueil and GRCB40 Universite´ Paul Sabatier- Toulouse III, Toulouse, France;3Michel Abbal, MD: Centre Hospi-
talier Universitaire Rangueil, Toulouse, France;4Arnaud Constantin,
MD, PhD: INSERM U558, Centre Hospitalier Universitaire Rangueil and GRCB40 Universite´ Paul Sabatier-Toulouse III, Toulouse, France.
Address correspondence and reprint requests to Arnaud Constantin, MD, PhD, Service de Rhumatologie, CHU Rangueil, 31059 Toulouse Cedex 9, France. E-mail: constant@cict.fr.
Submitted for publication August 17, 2005; accepted in revised form November 7, 2005.
In a recent study, Tezenas du Montcel et al proposed a new classification of HLA–DRB1 alleles that reconsiders the SE hypothesis of RA suscepti- bility (6,7). In addition to the RAA pattern at position 72–74, it takes into account the possible influence of amino acids at positions 70 and 71. Classification of the HLA–DRB1 alleles is thus achieved according to their amino acid sequence at position 70–74. Accord- ing to this new classification, susceptibility to RA de- pends on the RAA amino acid pattern at position 72–74 but is modulated by the amino acids at positions 70 and 71 (6).
In the present prospective longitudinal study, we investigated the relevance of the new classification pro- posed by Tezenas du Montcel et al in comparison with the classification previously described by Gao et al in terms of structural severity, in a cohort of French Caucasoid patients with early RA. Interestingly, this new classification of HLA–DRB1 alleles allows the differen- tiation between predisposing and protective alleles for RA severity.
PATIENTS AND METHODS
Patients. One hundred forty-four Caucasoid patients
(109 women and 35 men) were selected from the Rangueil Midi-Pyre´ne´es cohort, which comprised 826 patients with early arthritis who attended the Rangueil Hospital Department of Rheumatology between November 1992 and December 2002. The Disease Activity Score in 28 joints (DAS28) was used to quantify disease activity in patients with RA (8). Selection criteria included a followup period of at least 4 years (selection was restricted to patients included between No- vember 1992 and December 1997), the American College of Rheumatology (formerly, the American Rheumatism Asso- ciation) 1987 criteria for RA (9), disease duration⬍1 year from the first clinical manifestation of RA, and age over 16 years.
All patients included in the Rangueil Midi-Pyre´ne´es cohort signed an informed consent form, which was obtained after providing a verbal explanation of the research project and what their involvement would entail. The protocol was initially approved by the Committee for the Protection of Persons Participating in Biomedical Research (French law 88-1138; December 20, 1988).
Structural severity quantification. Radiographs of the
hands and feet were obtained at the start of the study and every year thereafter in all patients. All radiographs were scored by the same investigator (AC) according to the Sharp/van der Heijde method (10). The progression of the total radiographic damage score was calculated by subtracting the baseline radio-
graphic damage score from the 4-year followup radiographic damage score for each patient with RA. Progression of the total radiographic damage score was used to quantify the structural severity of RA.
HLA–DRB1 genotyping. Genomic DNA was ex-
tracted from EDTA-anticoagulated peripheral blood, using a standard proteinase K digestion and phenol–chloroform extraction method. HLA–DRB1 typing and subtyping were performed by a polymerase chain reaction–based method, using a panel of sequence-specific oligonucleotide probes (11).
HLA–DRB1 allele classifications. The first classifica-
tion pooled the HLA–DRB1 alleles according to the method proposed by Gao et al (5). Briefly, the HLA–DRB1*0101, *0405, *0408, *1001, and *1402 alleles were designated as E1, the HLA–DRB1*0401 alleles were designated as E2, the HLA–DRB1*0102 and *0404 alleles were designated as E3, and the other HLA–DRB1 alleles were designated as EX(5) (Table 1).
The second classification pooled the HLA–DRB1 alleles according to the new method proposed by Tezenas du Montcel and colleagues (6). Briefly, the HLA–DRB1 alleles were first divided into 2 groups according to the presence or absence of the RAA sequence at position 72– 74 and were denoted S and X alleles, respectively. The S alleles were subsequently divided into 4 groups according to the amino acid at position 71: an alanine (A), a glutamic acid (E), a lysine (K), or an arginine (R). Four different groups were thus defined in the new classification: S1 for ARAA and ERAA, S2for KRAA, S3for RRAA, and X for all non-RAA patterns (see Table 1). Moreover, it has been suggested that the amino acid at position 70 might influence susceptibility to RA (12); indeed, an aspartic acid (D) is supposed to be protective compared with the presence of a glutamine (Q) or an arginine (R). Thus, 2 additional groups were defined: S3D for DRRAA and S3P for QRRAA or RRRAA (6). The classifications introduced by Gao et al (5) and Tezenas du Montcel et al (6) are jointly presented in Table 1.
Statistical analysis. Due to the non-Gaussian distri-
bution of the total radiographic damage score, the median and interquartile range (IQR) were used to describe its distribution, and nonparametric tests were used for statisti- cal analysis. Agreement with Hardy-Weinberg equilibrium was tested using Pearson’s chi-square test. The progression of the total radiographic damage score was compared between the different groups of patients carrying or not carrying the HLA–DRB1 alleles, pooled according to the classifications described by Gao et al (5) and Tezenas du Montcel et al (6) using Wilcoxon’s rank sum (Mann- Whitney) test. The dose effect was investigated for alleles positively or negatively associated with structural severity, using a nonparametric trend test as an extension of Wilcox- on’s rank sum test. All computations were performed using Stata version 7.0 SE software (Stata, College Station, TX). AllP values were 2-sided, and P values less than 0.05 were considered significant.
RESULTS
Characteristics of RA patients at baseline and after 4 years of followup. Table 2 shows the main
characteristics of the RA patients (109 women and 35 men) at baseline and after 4 years of followup. Over the
4-year followup period, the mean ⫾ SD DAS28 de-
creased from 4.87⫾ 1.46 to 3.35 ⫾ 1.62 (P ⬍ 0.0001 by Wilcoxon’s matched pairs signed rank test), while the median total radiographic damage score increased from
0 (IQR 0–4) to 13 (IQR 3–35.5) (P ⬍ 0.0001 by
Wilcoxon’s matched pairs signed rank test). Over the 4 years of followup, the median progression of the total radiographic damage score was 11 (IQR 2–33.5) (P ⬍ 0.0001 by Wilcoxon’s matched pairs signed rank test).
Progression of structural severity according to the classification proposed by Gao et al. Based on the
classification described by Gao et al (5), we observed the following mean⫾ SD allele frequencies: 21.9 ⫾ 2.3% for E1 (63 of 288 alleles), 21.5 ⫾ 2.4% for E2 (62 of 288
alleles), 6.2⫾ 1.5% for E3(18 of 288 alleles), and 50.4⫾
Table 1. HLA–DRB1 amino acid sequence for alleles observed among patient groups, and their classification according to Gao et al and Tezenas du Montcel et al*
HLA–DRB1 alleles
Amino acid position
Gao classification Tezenas du Montcel classification 69 70 71 72 73 74 75 76 – 85 86 87 DRB1*0101 E Q R R A A V D V G E E1 S3P DRB1*0102 – Q R R A A – – A V – E3 S3P DRB1*0103 – D E R A A – – – – – Ex S1 DRB1*03 – – K – G R – – – V – Ex X DRB1*0401 – Q K R A A – – – – – E2 S2 DRB1*0402 – D E R A A – – – V – Ex S1 DRB1*0403 – – – – – E – – – V – Ex X DRB1*0404 – Q R R A A – – – V – E3 S3P DRB1*0405 – Q R R A A – – – V – E1 S3P DRB1*0407 – – – – – E – – – – – EX X DRB1*0408 – Q R R A A – – – G – E1 S3P DRB1*0411 – – – – – E – – – V – EX X DRB1*07 – D – – G Q – – – – – EX X DRB1*08 – D – – – L – – – – – EX X DRB1*0901 – R – – – E – – – – – EX X DRB1*1001 – Q R R A A – – – G – E1 S3P DRB1*1101 – D R R A A – – – – – EX S3D DRB1*1102 – D E R A A – – – V – EX S1 DRB1*1103 – D E R A A – – – V – EX S1 DRB1*1104 – D R R A A – – – V – EX S3D DRB1*12 – D R R A A – – A V – EX S3D DRB1*1301 – D E R A A – – – V – EX S1 DRB1*1302 – D E R A A – – – – – EX S1 DRB1*1303 – D K R A A – – – – – EX S2 DRB1*1323 – D E R A A – – – – – EX S1 DRB1*1401 – R – – – E – – – V – EX X DRB1*1402 – Q R R A A – – – G – E1 S3P DRB1*1404 – R – – – E – – – V – EX X DRB1*15 – D E R A A – – – V – EX S1 DRB1*16 – D R R A A – – – – – EX S3D
* The first classification clustered the HLA–DRB1 alleles according to the method proposed by Gao et al (5), as reported by Tezenas du Montcel et al (6). According to the classification by Gao et al, we designated the HLA–DRB1*0101, *0405, *0408, *1001, and *1402 alleles as E1, the
HLA–DRB1*0401 alleles as E2, the HLA–DRB1*0102 and *0404 alleles as E3, and the other HLA–DRB1 alleles as EX. This classification
differentiates the alleles known to be associated with rheumatoid arthritis according to the amino acids at positions 71 and 86 (underlined). In the Tezenas du Montcel et al classification, the HLA–DRB1 alleles were first divided into 2 groups according to the presence or absence of the RAA sequence at positions 72–74 (italics), which denotes S and X alleles, respectively. The S alleles were subsequently divided into 4 groups according to the 2 first amino acids at positions 70 and 71 (boldface): S1for ARAA and ERAA, S2for KRAA, S3for RRAA (divided into S3Pfor QRRAA
and S3Dfor DRRAA according to position 70), and X for all non-RAA motifs. The conventional classification of the amino acids was used, here
divided into 3 biochemical subgroups, as follows: group 1⫽ G for glycine, A for alanine, V for valine, L for leucine (aliphatic amino acids [nonpolar hydrophobic]); group 2⫽ K for lysine, R for arginine (basic amino acids [polar and positively charged]); group 3 ⫽ E for glutamic acid, Q for glutamine (the amide corresponding to E), D for aspartic acid, and N for asparagine (the amide corresponding to D) (acidic amino acids and corresponding amides are very hydrophilic; acidic amino acids polar and negatively charged at physiologic pH, amides polar and uncharged, and not ionizable).
3% for EX(145 of 288 alleles). No significant departures
from Hardy-Weinberg equilibrium were observed (Pear- son’s2⫽ 11.28, P ⫽ 0.08).
Table 3 shows the progression of structural se- verity over the 4-year followup period among patients carrying the different alleles, according to the classifica- tion proposed by Gao et al (5). The progression of the total radiographic damage score did not differ between patients carrying E1 alleles and noncarriers (P ⫽
0.8115). Patients carrying E2 or E3 alleles had greater
progression of the total radiographic damage score than did noncarriers (P ⫽ 0.004 and P ⫽ 0.0008, respectively). A significant dose effect was shown for E2and E3alleles
(P ⫽ 0.01 and P ⬍ 0.01, respectively). Based on the
classification proposed by Gao et al (5), both E2and E3
alleles appeared to be predisposing alleles for the sever- ity of joint damage in RA.
Table 4 presents the distributions for progression of structural severity for the different genotypes, accord- ing to the classification suggested by Gao et al (5). These data confirmed, at the genotype level, the predisposing effect of E2and E3that was demonstrated in the carrier
analysis and suggested that the predisposing effect of E3
is stronger than that of E2.
Table 5. Progression of structural severity over the 4-year followup period according to individual carrier status of the alleles, based on the Tezenas du Montcel classification*
HLA–DRB1 allele No. (%) of patients Progression of radiographic damage score, median (IQR) P† P for trend‡ S1(⫺) 95 (66) 11 (2–42) 0.4239 0.31 S1(⫹) 49 (34) 10 (1–20) S2(⫺) 88 (61.1) 9 (1–16.5) 0.0040 0.01 S2(⫹) 56 (38.9) 22 (5–46) S3P(⫺) 74 (51.4) 9 (1–28) 0.1585 0.15 S3P(⫹) 70 (48.6) 13 (3–37) S3D(⫺) 111 (77.1) 14 (4–42) ⬍0.0001 ⬍0.01 S3D(⫹) 33 (22.9) 2 (0–10) X (⫺) 95 (66) 12 (2–34) 0.4214 0.84 X (⫹) 49 (34) 10 (2–18)
* Radiographic damage (structural severity) was determined using the Sharp/van der Heijde method. IQR⫽ interquartile range.
† Carriers (⫹) versus noncarriers (⫺), by Wilcoxon’s signed rank test. ‡ By nonparametric test for trend across ordered groups.
Table 2. Characteristics of patients with RA at baseline and after 4 years of followup* Characteristic Baseline (n⫽ 144) 4-year followup (n⫽ 144) Age, mean⫾ SD years 48.8⫾ 13.7 52.9⫾ 13.7
No. of women/no. of men 109/35 –
Disease duration, mean⫾ SD months
6.73⫾ 3.7 56.09⫾ 4.6
DAS28, mean⫾ SD 4.87⫾ 1.46 3.35⫾ 1.62
Rheumatoid factor positive, no. (%)
99 (68.8) ND
Radiographic damage score, median (IQR)
0 (0–4) 13 (3–35.5)
* The Disease Activity Score in 28 joints (DAS28) ranges from 0 to 10. Scores⬎5.1 indicate high disease activity, and scores ⬍3.2 indicate low disease activity. The total radiographic damage score ranges from 0 to 448 and was quantified according to the Sharp/van der Heijde method. RA ⫽ rheumatoid arthritis; ND ⫽ not done; IQR ⫽ interquartile range.
Table 3. Progression of structural severity over the 4-year followup period according to individual carrier status of alleles, based on the Gao classification* HLA–DRB1 allele No. (%) of patients Progression of radiographic damage score,
median (IQR) P† P for trend‡
E1(⫺) 85 (59) 11 (1–37) 0.8115 0.57 E1(⫹) 59 (41) 10 (3–28) E2(⫺) 88 (61.1) 9 (1–16.5) 0.0040 0.01 E2(⫹) 56 (38.9) 22 (5–46) E3(⫺) 128 (88.9) 9 (1–27) 0.0008 ⬍0.01 E3(⫹) 16 (11.1) 37 (18.5–64.5) EX(⫺) 38 (26.4) 28.5 (7–61) 0.0006 ⬍0.01 EX(⫹) 106 (73.6) 8.5 (1–18)
* Radiographic damage (structural severity) was determined using the Sharp/van der Heijde method. IQR⫽ interquartile range.
† Carriers (⫹) versus noncarriers (⫺), by Wilcoxon’s signed rank test. ‡ By nonparametric testing for trend across ordered groups.
Table 4. Progression of structural severity over the 4-year followup period according to genotype, based on the Gao classification*
HLA–DRB1 genotype No. (%) of patients Genotype frequency, % Progression of radiographic damage score, median (IQR)† E3/E3 2 1.39 45 (26–64) E3/E1 5 3.47 17 (15–61) E3/E2 6 4.17 41 (37–66) E3/EX 3 2.08 20 (20–88) E2/E2 6 4.17 18 (8–42) E2/E1 15 10.42 29 (5–72) E2/EX 29 20.14 11 (2–34) E1/E1 4 2.78 0.5 (0–42) E1/EX 35 24.31 8 (2–14) EX/EX 39 27.08 6 (0–15)
* Radiographic damage (structural severity) was determined using the Sharp/van der Heijde method. Genotypes were sorted by strong-effect allele carriage. Homozygous genotypes appear in boldface.
† Kruskal-Wallis analysis was used to test the equal distribution of the progression of radiographic damage scores across all genotypes.P ⫽ 0.0045 by Kruskal-Wallis test, without adjustment for multiple testing. IQR⫽ interquartile range.
Progression of the structural severity according to the classification proposed by Tezenas du Montcel et al. Based on the classification suggested by Tezenas du
Montcel et al (6), we observed the following mean⫾ SD allele frequencies: 17.7⫾ 2.1% for S1(51 of 288 alleles),
21.5⫾ 2.4% for S2(62 of 288 alleles), 13.2⫾ 2.1% for
S3D(38 of 288 alleles), 28.1⫾ 2.6% for S3P (81 of 288
alleles), and 19.4 ⫾ 2.4% for X (56 of 288 alleles). No significant departures from Hardy-Weinberg equili- brium were observed (Pearson’s2⫽ 9.212, P ⫽ 0.51). Table 5 shows the progression of the total radio- graphic damage score over the 4-year followup period among patients carrying the different alleles, according to the classification proposed by Tezenas du Montcel et al (6). The progression of the total radiographic damage score did not differ between patients carrying S1or S3P
alleles and those who did not carry these alleles. Patients carrying S2 alleles had greater progression of the total
radiographic damage score than did noncarriers (P ⫽ 0.004). In contrast, patients carrying S3Dalleles had less
progression of the total radiographic damage score than did noncarriers (P ⬍ 0.0001). A significant dose effect was shown for S2and S3Dalleles (P ⫽ 0.01 and P ⬍ 0.01,
respectively). Based on the classification proposed by Tezenas du Montcel et al (6), S2alleles appeared to be
predisposing and S3D alleles appeared to be protective
for RA severity.
Table 6 presents the distribution of the progres- sion of structural severity for the different genotypes, according to the classification described by Tezenas du Montcel et al (6). These data confirm, at the genotype level, the protective allele effect of S3Dand the predis-
posing allele effect of S2that were demonstrated in the carrier analysis and suggest that the S3Dprotective effect
is stronger than the S2predisposing effect.
DISCUSSION
The results of the present prospective longitudi- nal study of patients with early RA support the previ- ously reported association between HLA–DRB1 alleles and the structural severity of RA. The classification suggested by Tezenas du Montcel et al does fit with linkage and association data for RA susceptibility (6) and appears to be relevant for RA severity in the present association study. While the classification of HLA– DRB1 alleles by Gao et al (5) identifies predisposing alleles for RA severity, the new classification proposed by Tezenas du Montcel et al allows the differentiation between predisposing and protective alleles for RA severity, stressing the interest of classifications of HLA– DRB1 alleles in the analysis of the major histocompat- ibility complex genetics of RA.
Both classifications allow the identification of predisposing alleles for RA severity: E2 and E3in the system described by Gao et al (5) and S2in the system
proposed by Tezenas du Montcel et al (6). E2 and S2
alleles overlap in the two classifications, corresponding to the HLA–DRB1*0401 allele, characterized by the KRAA amino acid pattern at position 71–74 (Table 1). In the present study, S2 perfectly matched E2, because
the HLA–DRB1*1303 allele (DKRAA pattern at posi- tion 70–74) was not observed in our sample. The partic- ular importance of the HLA–DRB1*0401 allele was recently emphasized in a meta-analysis that assessed the association of HLA–DRB1*SE alleles with bony ero- sions in RA. In that study, the HLA–DRB1*0401 allele appeared as the highest predisposing HLA–DRB1 allele for erosive RA among northern European Caucasoids (13). E3alleles that pool HLA–DRB1*0102 and *0404
alleles (QRRAA pattern at position 70–74) according to the classification suggested by Gao et al (5) appeared as predisposing alleles for structural severity in the present study. These results are consistent with the previously published results among northern European Cauca-