|
|
| |||||||||||||||||||||||||||||||
Background Rheumatoid arthritis frequently remits during pregnancy, for unknown reasons. Since an immune response to paternally inherited fetal HLA can occur during normal pregnancy and since rheumatoid arthritis is an autoimmune disorder with a known HLA class II antigen association, we tested the hypothesis that maternal-fetal disparity in HLA alloantigens might be associated with the pregnancy-induced remission of rheumatoid arthritis.
Methods We studied 57 pregnancies of 41 women with rheumatoid arthritis, 18 prospectively and 39 retrospectively. Serologic and DNA techniques were used to study HLA class I and II antigens. For newborns, typing was performed from cord-blood samples obtained at delivery. For four young children, typing was performed from DNA extracted from hair samples.
Results We found significantly more maternal-fetal disparity in HLA-DR and DQ antigens in pregnancies characterized by the remission or improvement of rheumatoid arthritis than in pregnancies characterized by active disease. Further studies using DNA-typing techniques to define allelic variants of HLA-DR and DQ antigens confirmed this observation. Maternal-fetal disparity in alleles of HLA-DRB1, DQA, and DQB occurred in 26 of 34 pregnancies characterized by remission or improvement (76 percent), as compared with 3 of 12 pregnancies characterized by active arthritis (25 percent) (odds ratio, 9.7; P = 0.003). The difference between the two groups was most marked for alleles of HLA-DQA.
Conclusions Amelioration of rheumatoid arthritis during pregnancy is associated with a disparity in HLA class II antigens between mother and fetus. These findings suggest that the maternal immune response to paternal HLA antigens may have a role in the pregnancy-induced remission of rheumatoid arthritis.
HLA class II antigen associations have been described in many autoimmune diseases, including rheumatoid arthritis6. Why the semi-allogeneic fetus escapes rejection is an intriguing unsolved question in immunology. Pregnancy presents an immunologic challenge, and maternal immune responses to paternally inherited fetal HLA antigens occur during normal pregnancy7. Because of these observations, we tested the hypothesis that a disparity between mother and fetus in HLA class II antigens might be associated with the remission of rheumatoid arthritis during pregnancy. We determined the HLA antigens of mothers and their children and analyzed the results with respect to whether the mothers' arthritis improved during pregnancy.
Methods
Patients
We studied 57 pregnancies (39 retrospectively and 18 prospectively) in 41 women who had rheumatoid arthritis according to the 1987 revised criteria of the American Rheumatism Association8. Thirty-one of the 38 women who had undergone radiography (82 percent) had erosive changes typical of rheumatoid arthritis. In 30 of the 40 women for whom the results of tests for rheumatoid factor were available (75 percent) the results were positive. Three women for whom test results were not available were tested at study entry, when their arthritis was inactive; all were negative for rheumatoid factor. These women might have tested positive earlier in the course of their illness, since they had had severe arthritis. All study pregnancies resulted in live births. The study was approved by the institutional review committee, and informed consent was obtained from all the subjects.
Assessment of Disease Activity
Disease activity was assessed according to the patients' symptoms and use of arthritis medication and by physical examination. The women were asked to judge the activity of their arthritis on a scale from 0 (inactive) to 3 (severely active) month by month for the six months before they became pregnant, during the pregnancy, and for three months after they gave birth. In 32 of the 39 pregnancies that were studied retrospectively (82 percent), we reviewed medical records for the physicians' assessments of disease activity. The women who were studied prospectively were examined for joint pain and swelling every two to three months, and a record was kept of morning stiffness and medications taken for symptoms of arthritis.
To be included in the groups having remission (group 1A) or improvement (group 1B) during pregnancy, a woman had to have had active arthritis in the six months before pregnancy. The first four of the six criteria for remission proposed by the American Rheumatism Association9 had to be met for inclusion in group 1A, including no use of arthritis medication, no more than 15 minutes of morning stiffness each day, no joint tenderness, and no joint swelling on physical examination. The two other criteria (absence of fatigue and an erythrocyte sedimentation rate of less than 30 mm per hour) were not included in our criteria for remission, because fatigue was not thought to be a reliable symptom of arthritis in a pregnant woman and because the erythrocyte sedimentation rate rises in normal pregnant women. Our criteria for remission also required that remission be present throughout at least the final trimester of pregnancy. Once achieved, remission was sustained in all pregnancies. Twenty-two pregnancies of 21 women were included in group 1A.
The criteria for improvement (group 1B) were a reduction or discontinuation of medication, no more than 15 minutes of morning stiffness each day, improvement in joint symptoms as compared with the situation in the six months before pregnancy, and no evidence of joint swelling during the third trimester. Most women we considered to have improved satisfied all four of these criteria; those satisfying three of the four included one woman whose symptoms were the same as before pregnancy and one woman who had morning stiffness for more than 15 minutes each day in the second and third trimesters. Three women with improvement (group 1B) had more symptoms during the second trimester than during the first trimester (Patients 13, 32, and 40) and one woman (Patient 47) had palindromic symptoms during the third trimester. (This woman had also had palindromic symptoms when rheumatoid arthritis developed.) Eleven women with 12 pregnancies were included in group 1B.
Group 2 consisted of women whose rheumatoid arthritis was inactive during pregnancy but for whom it was not certain whether the disease had been active in the six months before conception. Because disease that was in remission before pregnancy and remained in remission during pregnancy might differ from active disease that remitted during pregnancy, these pregnancies were not considered in our calculations of statistical significance and were not studied with DNA typing. Group 2 included four women with 11 pregnancies.
Group 3 consisted of women who had both subjective and objective (swelling on examination) evidence of active disease during pregnancy. This group included two women (Patients 13 and 28) whose first symptoms or diagnosis was during pregnancy. Group 3 included nine women with 12 pregnancies.
In 13 of the 46 pregnancies in groups 1 and 3 (28 percent), the women had taken disease-modifying medications in the six months before pregnancy (gold, sulfasalazine, hydroxychloroquine, or methotrexate). These drugs were discontinued before pregnancy or during the first trimester, when the woman became aware of the pregnancy.
HLA Typing
Women and their children underwent HLA typing with serologic methods and sequence-specific oligonucleotide probes. Fathers were included when available (85 percent of the pregnancies in groups 1 and 3) to confirm results and to assist in the assignment of haplotypes. HLA typing of newborn infants was performed with lymphocytes isolated from cord blood at the time of delivery. In four young children, typing was performed from DNA extracted from hair samples10.
Serologic techniques detect HLA glycoprotein molecules expressed on the cell surface; the term "HLA antigen" is used to refer to these molecules. "Alloantigen" is used to refer to an HLA antigen capable of stimulating an immune response in an HLA-nonidentical subject of the same species. Each HLA molecule is composed of two polypeptide chains; in class II molecules these chains are referred to as alpha and beta chains. The DRA, DRB, DQA, and DQB genes encode the DRalpha, DRbeta, DQalpha, and DQbeta chains, respectively. Alternative forms of genes at a given genetic locus are termed alleles. DRB1, DQA, and DQB genes are polymorphic, and DNA-based sequence-specific oligonucleotide probes are used to identify the alleles for these loci.
A standard National Institutes of Health two-stage complement-dependent microcytotoxicity assay was used to define HLA-A, B, C, DR, and DQ antigens11.
Typing with Sequence-Specific Oligonucleotide Probes
DNA was prepared from peripheral-blood leukocytes, and typing with sequence-specific oligonucleotide probes for HLA-DRB1 was performed with previously described methods12,13. A panel of 11 probes was developed to detect eight alleles of HLA-DQA, including DQA*0101, *0102, *0103, *0201, *0301, *0401, *0501, and *0601; the primers were those described by Scharf et al14. HLA-DQA probes are named for amino acid polymorphisms and positions (except 27NC, a nucleotide difference not resulting in an amino acid difference) and include (by name and specificity) 52S, *0101, *0102, and *0103; 34Q, *0102, *0103, and *0501; 41K, *0103; 52H, *0201; 26S, *0301; 51L, *0401, *0501, and *0601; 75S, *0501; 25F, *0601, *0201, and *0103; 34E, *0101; 25Y, *0101, *0102, *0401, and *0501; and 75IL, *0401, *0601, and *0201. A panel of 13 probes was used to detect 13 alleles of HLA-DQB, including DQB1*0501, *0502, *0503, *0601, *0602, *0603, *0604, *0201, *0301, *0302, *0303, *0401, and *0402. The DQB probe panel was modeled on a system and used the primers described by Morel et al15. The probes were 57S, *0502; 57D, *0602 and *0603; 37D, *0601; 27NC, *0603 and *0604; 57V, *0501 and *0604; 37I, *0201; 45E, *0301; 57PD, *0301 and *0303; 57PA, *0302; 57LD, *0401 and *0402; 26GH, *0501, *0502, and *0503; 45G, *0503, *0601, *0302, *0303, *0401, and *0402; and 23L, *0401. (The sequences of the probes are available elsewhere.*) The typing systems using the HLA-DQA and DQB sequence-specific oligonucleotide probes were validated with a panel of cells for which sequencing data have been published and cells that have been studied extensively as part of international HLA workshops and for which the DQ alleles are known. The probe panels distinguish 36 possible homozygous and heterozygous combinations of HLA-DQA and all except 2 of 91 possible homozygous and heterozygous combinations of DQB.
Statistical Analysis
Twins (Patient 25) were considered as one person in the statistical analysis. The 34 pregnancies in group 1 were compared with the 12 pregnancies in group 3. The statistical methods described by Zeger and Liang16 were used because eight women in the study had two pregnancies each. Although each possible pattern of disease activity in two pregnancies in the same woman was observed, these statistical methods address the potential problem of repeated pregnancies in the same woman and incorporate a correction for the possible correlation of repeated events in a single subject. In the secondary analyses for each of the three loci, HLA-DRB1, DQA, and DQB, the Bonferroni method17 was used to correct for multiple comparisons, and statistical significance was indicated by P<0.017 rather than P<0.05.
Results
Among the 57 pregnancies in 41 women with rheumatoid arthritis, 18 were studied prospectively and 39 retrospectively. Remission of disease occurred during 22 pregnancies (group 1A), and improvement during 12 (group 1B). Arthritis was active during 12 pregnancies (group 3). There was no difference between group 1A and group 1B in the extent of HLA disparities between mother and fetus, so these two groups were combined in the analyses. The maternal-fetal disparity in HLA antigens in group 1 as compared with group 3 is shown in Table 1. Disparity was examined from the perspective of the mother; it was considered to be present if a fetal HLA antigen differed from both maternal antigens, and it was considered to be absent if the fetal HLA antigen was the same as either maternal antigen. The frequency of maternal-fetal disparity in HLA-A and B antigens (class I) was similar in group 1 and group 3 (HLA-C antigens also did not differ [data not shown]). In contrast, there were more disparities between maternal and fetal class II antigens in group 1 than in group 3. There was a disparity in both HLA-DR and DQ in 76 percent of the pregnancies in group 1, as compared with 25 percent of the pregnancies in group 3 (odds ratio, 9.7; P = 0.003). There was no significant difference between the results for HLA-DR53 (DRB4) or HLA-DR52 (DRB3) in the two groups (data not shown). The results for group 2 were more similar to those for group 1 than to those for group 3, with 57 percent of the pregnancies involving disparity in HLA-DR and DQ; these results were not further analyzed, as previously explained.
|
|
|
|
Because of linkage disequilibrium in class II genes, the number of maternal-fetal pairs differing in only one locus was small. In just one pregnancy (Patient 47) was there isolated sharing of HLA-DQB between mother and fetus, with disparity in DRB1 and DQA. Arthritis improved in this woman, the only subject who had palindromic symptoms in the third trimester. Of the pregnancies in group 1, 12 percent involved maternal-fetal disparity in HLA-DRB1 but not DQA, as compared with 42 percent of the pregnancies in group 3. These results (in addition to the results of the secondary analysis) suggest that disparity only in HLA-DRB1 is insufficient for a favorable effect of pregnancy on rheumatoid arthritis. There was no disparity in either HLA-DRB1 or HLA-DQA in 9 percent of the pregnancies in group 1 and 33 percent of the pregnancies in group 3. Nevertheless, since 79 percent of the pregnancies in group 1 had maternal-fetal disparity in both HLA-DRB1 and HLA-DQA, as compared with 25 percent of the pregnancies in group 3, and since no mother and fetus had a disparity in DQA without a disparity in DRB1 (owing to linkage disequilibrium), DQA disparity independent of DRB1 disparity cannot be assessed. As previously shown, most of the pregnancies during which arthritis remitted or improved were characterized by fetal-maternal disparity in all three loci.
Discussion
The amelioration of rheumatoid arthritis induced by pregnancy was first described more than 50 years ago1. Improvement has been attributed to increased cortisol secretion, to sex hormones, and to a pregnancy-specific alpha-glycoprotein, but it is not likely that any of these substances are responsible2,3,4,5,25,26,27. We questioned whether a maternal immune response to fetal HLA antigens inherited from the father might be related to the remission of disease activity during pregnancy. In studies comparing women whose arthritis improved during pregnancy with women with active arthritis, we found that a maternal-fetal disparity in HLA class II antigens was significantly correlated with the amelioration of arthritis during pregnancy.
Trophoblastic cells do not express classic HLA antigens28. Antibodies to HLA antigens are detected in the serum of pregnant women,29,30,31 however, so a pregnant woman is exposed to fetal HLA antigens, either because fetal cells escape into the maternal circulation or for some other reason. Although the explanation for our findings is not known, a maternal antibody response to paternal HLA antigens represents one possibility. Antibodies to class II antigens can abrogate or modulate disease in animal models of autoimmune disease, including myasthenia gravis,32 multiple sclerosis,33 and collagen-induced arthritis34. The antibodies used in the studies of animals were directed against class II self antigens. However, clinical improvement in patients with rheumatoid arthritis has resulted from the administration of preparations of IgG immunoglobulins that were eluted from pooled human placentas35,36 and that contained alloantibodies to HLA-DR and DQ37.
In the secondary analysis among the different loci, the association of improvement in arthritis with maternal-fetal disparity in class II antigens was greater for HLA-DQ than for DR. Recent reports have demonstrated immune suppression restricted by HLA-DQ,38,39,40 and another explanation for our results might involve the activation of suppressor mechanisms. Since there are a number of hypotheses about the pathogenesis of autoimmunity in patients with rheumatoid arthritis, however, our results could be explained in a number of ways41. For example, HLA class II molecules, in addition to presenting foreign antigens, are also known to present self peptides. In a recent study in which peptides were eluted from mouse class II molecules and sequenced, one of the peptides eluted from the mouse I-A molecule (analogous to human DQ) was a self peptide derived from the I-Ealpha molecule (analogous to human HLA-DRalpha)42. In patients with rheumatoid arthritis, peptides derived from HLA class II molecules of the fetus might thus compete with or displace maternal self peptides according to the hypothesis that a defect in the presentation or recognition of self peptides is involved in the pathogenesis of the disease. Whatever the mechanism of remission of arthritis during pregnancy, it is likely to be complex and multifactorial. In addition, although the findings of our secondary analysis regarding HLA-DQ are intriguing, it is not possible from current studies to draw conclusions about the effect of disparity at an individual locus independent of the effect of other class II loci. Indeed, the majority of pregnancies in which remission or improvement occurred were characterized by maternal-fetal disparity in HLA-DRB1, DQA, and DQB.
In conclusion, amelioration of rheumatoid arthritis during pregnancy occurred most often when fetal and maternal HLA class II antigens differed. Further understanding of this association may provide insight into the pathogenesis of rheumatoid arthritis and possible treatment strategies.
Supported by grants from the Western Washington Arthritis Foundation and the National Institutes of Health (R29 AR39282 and AR39153) and by a Senior Investigator Award from the National Arthritis Foundation.
We are indebted to the following rheumatologists and obstetricians who referred women for the study: John Baldwin, Deborah Cahill, Patrick Campbell, Frank Detraglia, John Dickson, Carin Dugowson, Gregory Gardner, Bruce Gilliland, Richard Jimenez, Michael Kennedy, James Lane, Eric Larson, John Lipani, Kenneth Mahlin, Mart Mannik, Peter Mohai, Mary Musselman, Francis Nardella, Richard Neiman, Maiz O'Conner, Ina Oppliger, Patricia Otto, Steven Overman, Scott Pollock, Eric Sasso, Peter Simkin, Wayne Tsuji, Carol Wallace, Mark Wener, Robert Willkens, and especially Jeffrey Carlin, David Wisner, and Wayne Wallis. We are indebted to Gary Longton, M.S., and Steve Self, Ph.D., for assistance with the statistical analysis, to Ms. Alison Sell for assistance in the preparation of the manuscript, and to Dr. Michael Lockshin, Dr. Daniel Geraghty, Dr. Thomas Cotner, and Dr. Patrick Beatty for review of the manuscript. The cooperation of study families, including patients, fathers, and especially young children who were willing to give blood for study purposes, is gratefully acknowledged.
Source Information
From the Human Immunogenetics Program, Division of Clinical Research, Fred Hutchinson Cancer Research Center, 1124 Columbia St., Seattle, WA 98104, where reprint requests should be addressed to Dr. Nelson.
References
| |||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
Previous
