AMERICAN JOURNAL OF REPRODUCTIVE IMMUNOLOGY 28:222-225 © 1992 MUNKSGAARD
Pregnancy and Rheumatoid Arthritis: An Overview TIM D. SPECTOR AND J.A.P. DA SILVA Department ofRheumatology, St. Bartholomew's Hospital, London
The literature concerning the relationship of rheumatoid arthritis (RA) to pregnancy is reviewed. The amelioration of RA during pregnancy is a complex process. No specific factor that causes amelioration and is amenable to therapeutic intervention has been identified. Either RA is associated with lower fertility/fecundity, or pregnancy induces a protective effect against the development of the disease. The biological mechanism of these findings is not established. (Am J Reprod Immunol. 1992; 28:222-225.) Key words: Rheumatoid arthritis, pregnancy nulliparity, fertility ABSfRACT:
INTRODUCTION
''All authors have now recognized the influence ofthe female generative.organs on the development of articular rheumatism" (Charcot, 1881). Evidence accumulated over the past two decades has reinforced what seemed to be a common belief one hundred years ago. The concept that rheumatoid arthritis (RA) has a strong hormonal basis is supported by the female to male incidence ratio of between 2 and 3:1; gender differences in incidence are greater before the age of and progressively decline in old age, while disease onset tends to gather around the menopause. Additional support comes from the effect of pregnancy both on existing disease and on its subsequent development, and finally the alterations and effect!' of sex hormones in rheumatoid patients. THE EFFECTS OF PREGNANCY ON ESTABLISHED RA
In 1890, Sir Archibald Garrod wrote: "It is curious to observe the occurrence of pregnancy during the course ofthe disease (arthritis) appears to exert opposite influences in different cases, in some cases accelerating the progress of the malady and in others acting as a temporary check on its development."! However, the first, and now classic, report of a series of patients who went into remission while pregnant was published by Hench in 1938. 2 Since then a number of case studies have been published (Table I). Over 75% of women with RA enter remission during pregnancy. Clinical effects usually start in the first trimester and develop further later on, resulting in improvement not only of pain and stiffness but also of objective signs of inflammation and disease activity. The majority of patients will be able to stop most of their medication. Unfortunately, this remission is shortlived and over 80% of patients will have relapsed by three months postdelivery, the disease assuming a degree of activity similar to the one before pregnancy. So far it has
Submitted for publication June 25, 1992: accepted July 6, 1992. Address reprint request to Tim D. Spector, Department of Rheumatology, St. Bartholomew's Hospital, Charterhouse Square, London EC 1M 6BQ.
not been possible to identify any predictors of response, except that the effect of one pregnancy tends to be reproduced in subsequent gestations. Overall, pregnancy results in an extraordinary rate of disease remissions, unparalleled by any of the available treatments for the disease. Investigation of the mechanisms involved is a promising field in the search for etiopathogenic clues and novel therapeutic approaches. Pregnancy is associated with a state of general immunosuppression indispensable for the survival ofthe fetal allograft.l" Although humoral immunity does not seem significantly changed during pregnancy, cell-mediated immunity is remarkably depressed as demonstrated by increased susceptibility to specific intracellular infections, prolonged skin graft survival, and reduced skin reaction to tuberculin. These observations correlate with in vitro findings of reduced reactivity ofperipherallymphocytes from pregnant women to viral antigens and mitogens. Many pregnancy-associated factors have been shown to have important immunosuppressive properties and to contribute to these changes. Pregnancy is associated with reduced numbers of circulatinf lymphocytes, especially of the helper-inducer subset, 1 and increases in factors derived from the fetus, like o:-fetoprotein and changes in the carbohydrate side-chains of IgG, all of which have been suggested to play a role in RA remission. The changes in IgG glycoforms revert in the immediate puerperium, coinciding with clinical relapse. Similar changes have been seen in mice with collagen-induced arthritis. 16 Pregnancy-associated O:2-g1ycoprotein (PAG) has well established immunosuppressive properties; levels ofPAG are elevated in 75% to 90% of pregnant women. Some authors have reported a strong negative association between PAG levels and RA activity during pre~ancyl3 but others have failed to find any correlation. 1 Possible additional contributions may be made by nonspecific suppression of activity, reduction of enzyme production, or maternal blocking antibodies. Naturally, the hormonal changes that occur during gestation have received particular attention. The high levels of chorionic gonadotropin, human placental lactogen, and prolactin all have the potential for significant immunosuppressive effects. IS The role of prolactin has been reinforced by results with murine collagen-induced arthritis, a model equally susceptible to the ameliorating effects of pregnancy, where treatment with bromocriptine results in 50% reduction in severity ofthe postpartum flare of the disease.P However, in humans, the timing of postrelapse has not shown a clear relationship to breast-feeding or lactation.!" Glucocorticoids, including cortisol, increase steadily during pregnancy and by the third trimester levels are about double those of the nonpregnant state.P" These levels are, however, insufficient to explain the effects on RA, and there is a poor correlation between their return to normal and the onset of relapse after delivery.
PREGNANCY AND RA
223
TABLE I. Effect of Pregnancy in Women with Rheumatoid Arthritis Author (ref)
% remission
Number of patients
Hench 1938 (2) Lewis-Faning 1950 (3) Torrent 1951 (4) Oka 1953 (5) Hargreaves 1958 (6) Smith 1960 (7) Betson 1964 (8) Morris 1969 (9) Neely 1977 (10) Ostensen 1983 (11) Ostensen 1983 (12) Unger 1983 (13) Total
22 22 15 93 11 12 24 17 20 31 10 14
91 95 80 75 91 75 54 82 63 75 90 71
291
77
Sex steroids show remarkable increases during gestation, reaching by the third trimester levels that ar~ about 25-fold higher for progesterone, 100-fold for estradiol, and 10- to 15-fold for testosterone in comparison to nonpregnant levels. Given their wide spectrum of immunological effects,21 it is tempting to hypothesize that they may playa dominant role in this process, although their re~ ative importance is unclear. In collagen-induced arthritis in mice, treatment with estradiol has been shown to reduce the susceptibility to and the severity ofthe experimental arthritis. Moreover, Mattsson et al. 22 demonstrated that the maintenance of pregnancy levels of estradiol protected the animals against the postdelivery flare. This would suggest a pivotal role of the elevated levels of estrogens in the amelioration ofRA. In humans, limited epidemiological data do not support a close relation between the reduction of estradiol levels or return of menstruation. Moreover, early studies aiming at improving RA by means of gestational levels of e~trogens were disappointing.F' A recent open study used high-dose combined contraceptive (Lyndiol) in ten rheumatoid women; only four patients completed the six-month stu~y period and significant improvement was only found III the number ofswollenjoints.i" The amelioration of RA during pregnancy is undoubtedly a complex process. Research has failed so far to isolate any specific mechanism or factor which may be dominant and open the way for therapeutic intervention.
% relapse at 3 months
77 81 84 100
62 100 81
Pregnancy and the Risk ofSubsequent Development ofRA Following early observations that women with RA tended to have smaller families than expected," a number of studies have examined the role of pregnancy in altering the subsequent risk of RA. These are shown in Table II. 25--34 Overall, the majority of reports show a fairly consistent increased risk ofRA in nUlli~arous women of approximately twofold. Only one study 1 demonstrated an effect varying with the number of pregnancies, while most authors find no increased protection with additional gestations. There is also no clear pattern that age at first pregnancy is strongly related to risk. The lack ofprotection found in the large prospective study in nurses 29 might suggest that pregnancy results in milder disease rather than an all or nothing effect, in the line of what has been suggested for the effects of oral contraceptives.i" Different groups have also shown that RA incidence is reduced during pregnancy but is increased more than expected in the 12 months postpartum. 5 ,30 ,34 A similar pattern has been reported in psoriatic arthritis.P" This seems to be a real, albeit unexplained, phenomenon unlikely to be explained by selection or recall bias. It is interesting to note that this period coincides with a rebound increase in IgG glycoforms following the depression during pregnancy which may reach pathological levels. 16 . The interpretation of these data in terms of pathogenesis and cause-effect relationship is not easy. Two main
TABLE II. NuUiparity and Risk of Rheumatoid Arthritis Author (ref)
Cases
Controls
Type of study Population survey
Engel 1962 (25) Kay 1965 (26) Lawrence 1977 (27) Vandenbroucke 1986 (28) Hernandez-Avila 1988 (29)a Del Junco 1989 (30)
209 267 148 115 324
209 (population) 1015 185 (OA + soft tissue) 121.700 324
Hazes 1990 (31) Koepsell 1990 (32) Spector 1990 (33)
135 101 270
Silman 1992 (34)
88
378 (OA + sfot tissue) 398 292 (OA) 245 (population) 144
Case-control Population survey Prospective nurses Population-based Case-control Case-control (incident) Case-control (incident) Case-control Case-control
RR for nulliparity (95% Cl, where given) 1.3 (age 35-44) 0.5 (age 45-54) 1'.97 1.8 No relation reported 3.16 2.04 (1.09-3.70) 2.0 1.82 (1.09-3.03) 1.83 (1.03-3.06) No relation reported
"Prospective study of121,700 female nurses totaling 883,187 person-years. New cases: 115 definite RA and 102 probable RA. RR = relative risk; OA = osteoarthritis.
224
SPECTOR AND DA SILVA
TABLE III. Studies of Prior Fertility and Reproductive Outcomes in RA Women Compared to Controls Author (ref) Kay 1965 (26)a Kaplan 1986 (37) Del Junco 1988 (30)a Siamapolou 1988 (38) Silman 1988 (39)b McHugh 1989 (40) Spector 1990 (41) Nelson 1990 (42)
Cases
Controls
Fertility (no. preg/person)
Spontaneous abortions
209 96 324 72 40 117 195 144
209 113 324 98 67 100 462 605
dec NS dec NS NS dec NS NS
NS inc NS NS NS NS dec NS
Stillbirth
Mean family size dec
inc NS NS
NS
'Without exclusion of nulliparous women. bComparison between RA patients and disease-free relatives. NS, no significant difference. inc, increase. dec, decrease.
possibilities are: either RA is associated with lower fertility/fecundity, or pregnancy induces, in some way, a real protective effect against the development ofthe disease. The increase in nulliparity among RA women might be explained by increased rates of reproductive loss antedating the disease. Studies addressing this are summarized in Table 111. 2 6 ,3 0 ,3 7-42 Only the study by Kaplan"? showed increased rates of miscarriage, but this might have resulted from .~thnic selection and poor matchin~. The higher rates of stillbirth reported by Silman et al. 9 probably result from an unexplained low rate among the relatives of patients. Overall it seems that RA is not associated with a higher incidence of spontaneous or induced abortions or of stillbirth, except perhaps as a result of severe established disease and aggressive therapy. Another possibility is that women predisposed to RA may have lower levels offertility or higher rates of infertility antedating, even by decades, the onset ofthe disease as seen with other autoimmune conditions.v'
Pregnancy and Rheumatoid Arthritis: An Overview TIM D. SPECTOR AND J.A.P. DA SILVA Department ofRheumatology, St. Bartholomew's Hospital, London
The literature concerning the relationship of rheumatoid arthritis (RA) to pregnancy is reviewed. The amelioration of RA during pregnancy is a complex process. No specific factor that causes amelioration and is amenable to therapeutic intervention has been identified. Either RA is associated with lower fertility/fecundity, or pregnancy induces a protective effect against the development of the disease. The biological mechanism of these findings is not established. (Am J Reprod Immunol. 1992; 28:222-225.) Key words: Rheumatoid arthritis, pregnancy nulliparity, fertility ABSfRACT:
INTRODUCTION
''All authors have now recognized the influence ofthe female generative.organs on the development of articular rheumatism" (Charcot, 1881). Evidence accumulated over the past two decades has reinforced what seemed to be a common belief one hundred years ago. The concept that rheumatoid arthritis (RA) has a strong hormonal basis is supported by the female to male incidence ratio of between 2 and 3:1; gender differences in incidence are greater before the age of and progressively decline in old age, while disease onset tends to gather around the menopause. Additional support comes from the effect of pregnancy both on existing disease and on its subsequent development, and finally the alterations and effect!' of sex hormones in rheumatoid patients. THE EFFECTS OF PREGNANCY ON ESTABLISHED RA
In 1890, Sir Archibald Garrod wrote: "It is curious to observe the occurrence of pregnancy during the course ofthe disease (arthritis) appears to exert opposite influences in different cases, in some cases accelerating the progress of the malady and in others acting as a temporary check on its development."! However, the first, and now classic, report of a series of patients who went into remission while pregnant was published by Hench in 1938. 2 Since then a number of case studies have been published (Table I). Over 75% of women with RA enter remission during pregnancy. Clinical effects usually start in the first trimester and develop further later on, resulting in improvement not only of pain and stiffness but also of objective signs of inflammation and disease activity. The majority of patients will be able to stop most of their medication. Unfortunately, this remission is shortlived and over 80% of patients will have relapsed by three months postdelivery, the disease assuming a degree of activity similar to the one before pregnancy. So far it has
Submitted for publication June 25, 1992: accepted July 6, 1992. Address reprint request to Tim D. Spector, Department of Rheumatology, St. Bartholomew's Hospital, Charterhouse Square, London EC 1M 6BQ.
not been possible to identify any predictors of response, except that the effect of one pregnancy tends to be reproduced in subsequent gestations. Overall, pregnancy results in an extraordinary rate of disease remissions, unparalleled by any of the available treatments for the disease. Investigation of the mechanisms involved is a promising field in the search for etiopathogenic clues and novel therapeutic approaches. Pregnancy is associated with a state of general immunosuppression indispensable for the survival ofthe fetal allograft.l" Although humoral immunity does not seem significantly changed during pregnancy, cell-mediated immunity is remarkably depressed as demonstrated by increased susceptibility to specific intracellular infections, prolonged skin graft survival, and reduced skin reaction to tuberculin. These observations correlate with in vitro findings of reduced reactivity ofperipherallymphocytes from pregnant women to viral antigens and mitogens. Many pregnancy-associated factors have been shown to have important immunosuppressive properties and to contribute to these changes. Pregnancy is associated with reduced numbers of circulatinf lymphocytes, especially of the helper-inducer subset, 1 and increases in factors derived from the fetus, like o:-fetoprotein and changes in the carbohydrate side-chains of IgG, all of which have been suggested to play a role in RA remission. The changes in IgG glycoforms revert in the immediate puerperium, coinciding with clinical relapse. Similar changes have been seen in mice with collagen-induced arthritis. 16 Pregnancy-associated O:2-g1ycoprotein (PAG) has well established immunosuppressive properties; levels ofPAG are elevated in 75% to 90% of pregnant women. Some authors have reported a strong negative association between PAG levels and RA activity during pre~ancyl3 but others have failed to find any correlation. 1 Possible additional contributions may be made by nonspecific suppression of activity, reduction of enzyme production, or maternal blocking antibodies. Naturally, the hormonal changes that occur during gestation have received particular attention. The high levels of chorionic gonadotropin, human placental lactogen, and prolactin all have the potential for significant immunosuppressive effects. IS The role of prolactin has been reinforced by results with murine collagen-induced arthritis, a model equally susceptible to the ameliorating effects of pregnancy, where treatment with bromocriptine results in 50% reduction in severity ofthe postpartum flare of the disease.P However, in humans, the timing of postrelapse has not shown a clear relationship to breast-feeding or lactation.!" Glucocorticoids, including cortisol, increase steadily during pregnancy and by the third trimester levels are about double those of the nonpregnant state.P" These levels are, however, insufficient to explain the effects on RA, and there is a poor correlation between their return to normal and the onset of relapse after delivery.
PREGNANCY AND RA
223
TABLE I. Effect of Pregnancy in Women with Rheumatoid Arthritis Author (ref)
% remission
Number of patients
Hench 1938 (2) Lewis-Faning 1950 (3) Torrent 1951 (4) Oka 1953 (5) Hargreaves 1958 (6) Smith 1960 (7) Betson 1964 (8) Morris 1969 (9) Neely 1977 (10) Ostensen 1983 (11) Ostensen 1983 (12) Unger 1983 (13) Total
22 22 15 93 11 12 24 17 20 31 10 14
91 95 80 75 91 75 54 82 63 75 90 71
291
77
Sex steroids show remarkable increases during gestation, reaching by the third trimester levels that ar~ about 25-fold higher for progesterone, 100-fold for estradiol, and 10- to 15-fold for testosterone in comparison to nonpregnant levels. Given their wide spectrum of immunological effects,21 it is tempting to hypothesize that they may playa dominant role in this process, although their re~ ative importance is unclear. In collagen-induced arthritis in mice, treatment with estradiol has been shown to reduce the susceptibility to and the severity ofthe experimental arthritis. Moreover, Mattsson et al. 22 demonstrated that the maintenance of pregnancy levels of estradiol protected the animals against the postdelivery flare. This would suggest a pivotal role of the elevated levels of estrogens in the amelioration ofRA. In humans, limited epidemiological data do not support a close relation between the reduction of estradiol levels or return of menstruation. Moreover, early studies aiming at improving RA by means of gestational levels of e~trogens were disappointing.F' A recent open study used high-dose combined contraceptive (Lyndiol) in ten rheumatoid women; only four patients completed the six-month stu~y period and significant improvement was only found III the number ofswollenjoints.i" The amelioration of RA during pregnancy is undoubtedly a complex process. Research has failed so far to isolate any specific mechanism or factor which may be dominant and open the way for therapeutic intervention.
% relapse at 3 months
77 81 84 100
62 100 81
Pregnancy and the Risk ofSubsequent Development ofRA Following early observations that women with RA tended to have smaller families than expected," a number of studies have examined the role of pregnancy in altering the subsequent risk of RA. These are shown in Table II. 25--34 Overall, the majority of reports show a fairly consistent increased risk ofRA in nUlli~arous women of approximately twofold. Only one study 1 demonstrated an effect varying with the number of pregnancies, while most authors find no increased protection with additional gestations. There is also no clear pattern that age at first pregnancy is strongly related to risk. The lack ofprotection found in the large prospective study in nurses 29 might suggest that pregnancy results in milder disease rather than an all or nothing effect, in the line of what has been suggested for the effects of oral contraceptives.i" Different groups have also shown that RA incidence is reduced during pregnancy but is increased more than expected in the 12 months postpartum. 5 ,30 ,34 A similar pattern has been reported in psoriatic arthritis.P" This seems to be a real, albeit unexplained, phenomenon unlikely to be explained by selection or recall bias. It is interesting to note that this period coincides with a rebound increase in IgG glycoforms following the depression during pregnancy which may reach pathological levels. 16 . The interpretation of these data in terms of pathogenesis and cause-effect relationship is not easy. Two main
TABLE II. NuUiparity and Risk of Rheumatoid Arthritis Author (ref)
Cases
Controls
Type of study Population survey
Engel 1962 (25) Kay 1965 (26) Lawrence 1977 (27) Vandenbroucke 1986 (28) Hernandez-Avila 1988 (29)a Del Junco 1989 (30)
209 267 148 115 324
209 (population) 1015 185 (OA + soft tissue) 121.700 324
Hazes 1990 (31) Koepsell 1990 (32) Spector 1990 (33)
135 101 270
Silman 1992 (34)
88
378 (OA + sfot tissue) 398 292 (OA) 245 (population) 144
Case-control Population survey Prospective nurses Population-based Case-control Case-control (incident) Case-control (incident) Case-control Case-control
RR for nulliparity (95% Cl, where given) 1.3 (age 35-44) 0.5 (age 45-54) 1'.97 1.8 No relation reported 3.16 2.04 (1.09-3.70) 2.0 1.82 (1.09-3.03) 1.83 (1.03-3.06) No relation reported
"Prospective study of121,700 female nurses totaling 883,187 person-years. New cases: 115 definite RA and 102 probable RA. RR = relative risk; OA = osteoarthritis.
224
SPECTOR AND DA SILVA
TABLE III. Studies of Prior Fertility and Reproductive Outcomes in RA Women Compared to Controls Author (ref) Kay 1965 (26)a Kaplan 1986 (37) Del Junco 1988 (30)a Siamapolou 1988 (38) Silman 1988 (39)b McHugh 1989 (40) Spector 1990 (41) Nelson 1990 (42)
Cases
Controls
Fertility (no. preg/person)
Spontaneous abortions
209 96 324 72 40 117 195 144
209 113 324 98 67 100 462 605
dec NS dec NS NS dec NS NS
NS inc NS NS NS NS dec NS
Stillbirth
Mean family size dec
inc NS NS
NS
'Without exclusion of nulliparous women. bComparison between RA patients and disease-free relatives. NS, no significant difference. inc, increase. dec, decrease.
possibilities are: either RA is associated with lower fertility/fecundity, or pregnancy induces, in some way, a real protective effect against the development ofthe disease. The increase in nulliparity among RA women might be explained by increased rates of reproductive loss antedating the disease. Studies addressing this are summarized in Table 111. 2 6 ,3 0 ,3 7-42 Only the study by Kaplan"? showed increased rates of miscarriage, but this might have resulted from .~thnic selection and poor matchin~. The higher rates of stillbirth reported by Silman et al. 9 probably result from an unexplained low rate among the relatives of patients. Overall it seems that RA is not associated with a higher incidence of spontaneous or induced abortions or of stillbirth, except perhaps as a result of severe established disease and aggressive therapy. Another possibility is that women predisposed to RA may have lower levels offertility or higher rates of infertility antedating, even by decades, the onset ofthe disease as seen with other autoimmune conditions.v'
