To analyze the variation of steroid hormone levels during pregnancy in patients with systemic lupus erythematosus (SLE). Moreover, to investigate whether, during gestation, there is any relationship between steroid concentration and SLE activity.
Seventeen consecutive pregnant SLE patients and 8 matched healthy pregnant controls were studied prospectively. Disease activity was evaluated by European Consensus Lupus Activity Measure (ECLAM) score modified for pregnancy. The following hormones were evaluated: testosterone, 17β-estradiol (estradiol), cortisol, dehydroepiandrosterone sulfate (DHEAS), and progesterone.
Disease activity score significantly varied during pregnancy and postpartum (P< 0.05), being decreased in the third trimester and increased in the second trimester and postpartum. Serum levels of all steroids varied significantly during pregnancy and the postpartum period both in patients and in healthy subjects. In SLE patients, estradiol, progesterone, and DHEAS concentrations were found to be significantly reduced compared with controls. Serum level profiles of estradiol and progesterone were different from those observed in controls. No differences in the steroid levels were observed between patients taking prednisone ≤5 mg/day or >5 mg/day, apart from cortisol, which was, as expected, lower in the latter group.
The major hormonal alteration observed during pregnancy in SLE patients was an unexpected lack of estrogen serum level increase, and, to a lesser extent, progesterone serum level increase, during the second and—even more—the third trimester of gestation. This lack of increase probably was due to placental compromise. Therefore, these steroid hormone variations may result in a lower humoral immune response activation, probably related to a change in the estrogen/androgen balance, that in turn could account for the decrease in disease activity observed during the third trimester in pregnant SLE patients.
The susceptibility to autoimmune diseases, such as rheumatoid arthritis (RA) and systemic lupus erythematosus (SLE), arises from complex interactions between several genetic, hormonal, and environmental predisposing factors. Steroid hormones play a relevant role because of their strong modulating effects on the immune system (1–9). Steroid hormones account for the higher immune reactivity and susceptibility to autoimmunity in women compared with men during the reproductive age. Estrogens seem to be mainly implicated as major enhancers of the immune response, whereas cortisol and androgens seem to act as natural suppressors (2–4).
In particular, a recent study confirms that 17β-estradiol is able to polyclonally increase the production of IgG, including IgG anti–double- stranded DNA, in SLE patients' peripheral blood mononuclear cells by enhancing B cell activity and by promoting interleukin-10 (IL-10) production by monocytes (5).
Pregnancy is a physiologic condition in which the sex hormone balance is modified to achieve immunosuppression and tolerance of the immune system to paternal and fetal antigens (10, 11). It is well known that estrogens progressively increase in the maternal circulation during pregnancy, particularly in the third trimester (12). Although less markedly, an increase in cortisol, dehydroepiandrosterone sulfate (DHEAS), progesterone, and testosterone are also observed during pregnancy.
Estrogen immunomodulation is dose dependent (6). At physiologic concentrations, estrogens seem to stimulate both humoral and cellular immune responses (Th2- and Th1-type cytokines). However, at supraphysiologic concentrations, as achieved during late pregnancy (at least 10-fold), estrogens seem to be immunosuppressive, at least on cell- mediated immune responses (Th1-type cytokines); but they induce antibody production (Th2-type cytokines). Even progesterone and glucocorticoids seem to suppress Th1 (10, 13–15) and increase Th2 cytokine production (14–17). The Th2 cytokine polarization could explain why RA patients generally improve (18) and SLE patients worsen during pregnancy (19). In fact, in SLE, pregnancy has long been considered a high-risk event, both for the mother and the fetus, because disease flares are frequently observed during gestation, and fetal wastage, prematurity, and low birth weight babies are more common in women with SLE than in the healthy population (20–24). In the last decades, however, it has become apparent that pregnancies resulted in favorable outcomes in the majority of cases, although SLE flares still remain common during pregnancy and postpartum (22, 25–32). Various recent observations suggest that the percentage of SLE flares falls in the third trimester of pregnancy (22, 27, 28, 31–34), although this is the period in which the highest serum levels of estrogen, progesterone, cortisol, and testosterone should be expected.
There are few studies regarding the relationships among steroid hormones, disease activity, and immunologic parameters during pregnancy in SLE (24). To clarify this important and poorly investigated issue, we prospectively analyzed the steroid hormone concentrations, including estrogen, progesterone, testosterone, DHEAS, and cortisol, in patients with SLE and matched healthy women before, during, and after pregnancy. Moreover, we investigated whether there was any relationship among steroid levels, disease activity, and some immunologic parameters.
PATIENTS AND METHODS
Seventeen consecutive successful pregnancies in 17 patients with SLE were prospectively studied. All patients (mean age 30.5 years, range 21–38; mean disease duration 91 months, range 18–184) fulfilled the American College of Rheumatology criteria for SLE (35). Eight pregnancies in 8 healthy volunteers (mean age 29 years, range 20–35) were considered as controls. Gestational period was similar in SLE patients (mean 37 weeks, range 31– 41) and controls (mean 38 weeks, range 35–41).
All patients were enrolled in our protocol for pregnancy planning and followup and they were evaluated monthly by the same rheumatologist and obstetrician during their entire pregnancy, including 10–12 weeks of the postpartum period.
Twelve of 17 pregnancies were fully planned and 5 pregnancies started unexpectedly. In our protocol, the pregnancy was planned when disease was inactive for at least 6 months.
At each clinical evaluation, routine laboratory tests were performed, including white blood cell count; urinalysis; and glucose, blood urea nitrogen, serum creatinine, IgG, IgM, and IgA levels. Antinuclear antibodies (ANA) and anti-DNA antibodies were detected by indirect immunofluorescence using as substrate HEp-2 cells and Crithidia luciliae, respectively. Anti–extractable nuclear antigen antibodies were detected by counterimmunoelectrophoresis; serum complement fractions C3 and C4 and C-reactive protein were detected by nephelometry. Anticardiolipin antibodies were detected by enzyme-linked immunosorbent assay (ELISA), and lupus anticoagulant was detected by Russell viper venom time assay.
SLE activity was measured by European Consensus Lupus Activity Measure (ECLAM) score (36), modified for pregnancy. We changed 3 of 12 ECLAM items as follows: in item number 9 we considered proteinuria, ≥ 500 mg/day, only after preeclampsia was ruled out; non-hemolitic anemia (item number 10) was not considered due to its high frequency during pregnancy; and, in item number 11, erythrocyte sedimentation rate (ESR) was not considered because it usually increases during pregnancy, mainly due to anemia and hemodilution.
The laboratory and clinical findings of our patients at conception are summarized in Table 1. For the purpose of this study, we considered only those pregnancies that ended in live birth. Data regarding pregnancy outcome are shown in Table 2.
Table 1. Clinical and laboratory findings of SLE patients at conception*
Serum samples for hormone testing were collected in the following periods: within the 3 months prior to pregnancy in the 12 planned pregnancies; at 9, 17, and 29 weeks of pregnancy; and at 1 month after delivery in all cases. All serum samples were taken at 8:00 AM and kept frozen until the hormone assays were performed.
Serum concentrations of cortisol (ng/ml), 17β-estradiol (estradiol; pg/ml), progesterone (ng/ml), and testosterone (ng/ml) were determined by chemiluminescence assay (Immulite, Diagnostic Product Corporation, Los Angeles, CA; sensitivity: cortisol 2 ng/ml, estradiol 10 pg/ml, progesterone 0.2 ng/ml, testosterone 0.1 ng/ml). DHEAS (μg/dl) was assayed by ELISA (Biochem Immunosystems Italia, Bologna, Italy; sensitivity: 5 μg/dl).
Programs from the BMDP statistical package (37) were used for calculations. A two-tailed analysis was used. The analysis of variance (ANOVA) for repeated measures was performed on protocol variables, considering the periods of data collection as within-subject factors and clinical classification as grouping factors. Within-group contrasts were performed according to Bonferroni's method. Results of ANOVA are presented reporting 3 F test values, namely FG for grouping factor, FT for time, and FGXT for the interaction between time and grouping factors. The BMDP program 6D was used for linear correlation and regression analysis.
Because most of the pregnancies (70%) were planned on the basis of inactive disease, the patient mean ECLAM score at conception was found to be substantially low (Table 1). However, the disease was moderately active at conception in 3 of 5 patients in whom pregnancy unexpectedly occurred (cases 4, 5, and 8; Table 1). Seven patients had a flare during pregnancy or postpartum (41%): 2 with unexpected and 5 with planned pregnancy. Five flares were observed in the second trimester and 2 in the postpartum period. Disease activity according to the ECLAM score varied significantly during pregnancy and postpartum (F = 2.88, P < 0.05), being higher in the second trimester and the postpartum period and lower in the third trimester (Figure 1). We also observed a significant variation of some hematologic and immunologic parameters during pregnancy and postpartum. Leukocyte count was higher, and γ globulins and ANA titer were lower in the third compared with the second trimester and postpartum (Table 3).
Table 3. Hematologic and immunologic parameters (mean ± SD) during pregnancy and postpartum in 17 SLE patients*
SLE = systemic lupus erythematosus; ANA = antinuclear antibodies.
Leukocytes (n × 109/ liter)
5.67 ± 1.78
6.72 ± 2.24
7.60 ± 2.52
5.72 ± 2.12
Lymphocytes (n × 109/ liter)
1.39 ± 0.50
1.34 ± 0.46
1.52 ± 0.58
1.72 ± 0.60
Gamma globulins (g/liter)
13.87 ± 3.92
11.51 ± 2.72
10.62 ± 2.99
13.71 ± 3.18
1,922 ± 2,481
1,668 ± 2,359
1,498 ± 2,381
1,536 ± 2,371
No patients had active renal disease at conception and all had normal renal function before pregnancy, throughout pregnancy, and postpartum. However, we observed proteinuria in 3 cases, all during the second trimester.
All but 3 patients were taking prednisone throughout pregnancy and postpartum. The dosage was maintained in those patients in whom the disease did not flare and was increased in those who relapsed, in order to control SLE activity. Thereafter, the dosage was soon tapered to the minimal possible dosage. Two patients were taking antimalarial drugs and one was taking cyclosporin A during the pregnancy. In one patient (number 10; Table 1), mild proteinuria was first diagnosed as preeclampsia and no therapy was introduced. Because proteinuria persisted for 1 year after delivery, the patient underwent renal biopsy that showed mesangial glomerulonephritis (World Health Organization class II).
Within the 3 months before pregnancy, among the steroid hormones tested, only DHEAS mean levels were significantly lower in SLE patients compared with healthy women (Table 4)
Table 4. Hormone levels (mean ± SD) before pregnancy in SLE patients and healthy controls*
Serum levels of all steroids varied significantly during pregnancy and postpartum, both in the patients and healthy controls, as expected (Figures 2–4): DHEAS (FT = 9.6, P = 0.002), testosterone (FT = 7.0, P = 0.0019), estradiol (FT = 46.4, P = 0.0001), progesterone (FT = 31.2, P = 0.0001), and cortisol (FT = 26.2, P = 0.0001).
Some noteworthy and relevant differences in steroid hormone levels and profiles between patients and healthy controls were observed. Total levels of DHEAS (Figure 2) the adrenal androgen with immunosuppressive activity, were significantly lower in patients compared with pregnant controls throughout gestation (FG = 9.7, P = 0.0049), although the hormone level profile was similar in both groups. Conversely, total serum testosterone concentration (the gonadal androgen with immunosuppressive activities) and its concentration profile during pregnancy did not differ between SLE patients and controls (Figure 2).
Estradiol level (Figure 3) showed the most relevant change in patients versus controls. Its concentration was significantly reduced in SLE patients, particularly in the third trimester of pregnancy (FG = 34.5, P < 0.0001), leading to a completely different profile compared with healthy pregnant subjects (FGXT = 12.5, P < 0.0001). The progesterone levels were found to be significantly reduced in patients compared with controls (FG = 11.3, P = 0.0027) and the hormone level profile (Figure 3) was also different (FGXT = 12.5, P < 0.0001). The total serum cortisol levels (Figure 4) were similar in both patients and controls; however, the hormone concentration during pregnancy varied significantly between the two groups (FGXT = 6.3, P = 0.0019).
To verify the possibility that sex hormone alterations might have been influenced by concomitant corticosteroid treatment, the patients were subdivided into 2 groups: the first group included 8 patients (3 of whom were steroid-free) who were taking a prednisone dosage of ≤5 mg/day (mean ± SD, 3 ± 2.5 mg/day), the second group included 9 patients who were taking >5 mg/day (mean ± SD, 16 ± 5.6 mg/day). A comparison of steroid hormone levels did not show any differences between SLE patients and controls, apart from cortisol concentration (Figure 4) that, as expected, was significantly reduced in SLE patients taking a prednisone dosage of >5 mg/day (FG = 4.9, P < 0.042). However, when SLE patients treated with prednisone at ≤5 mg/day (including 3 not treated) were compared with the control group, the differences for all hormone concentrations were less evident than between overall SLE patients and healthy women, as expected (Figures 2–4).
We did not observe any correlation between steroid hormone concentrations and ECLAM score, leukocyte and lymphocyte count, gamma globulin, IgG, IgA, IgM, and ANA titer, within any distinct gestational period.
Pregnancy is a condition during which the steroid hormone system undergoes profound changes, and is an optimum physiologic dynamic model to investigate the steroid hormone balance and steroid-induced immunomodulation in terms of clinical evolution of immune-mediated diseases.
Up to now, just one prospective study has been published concerning steroid hormone levels in relation to disease activity in pregnant women with SLE (38). In this study, prolactin was increased, whereas estrogens and testosterone were reduced throughout pregnancy in 9 SLE patients compared with 9 healthy and 5 RA subjects. These abnormalities seemed more relevant in patients with active disease. However, data regarding steroid hormone levels, and on progesterone, DHEAS, and cortisol levels before pregnancy and postpartum were lacking.
In our study, the steroid hormone setting before pregnancy did not differ in SLE compared with healthy subjects, apart from DHEAS serum levels, which were markedly decreased in SLE patients. According to some investigators (2), the DHEAS low level in SLE is due to adrenal hypofunction; alternatively, it could be the consequence of a DHEAS abnormal metabolism.
In addition, significant variations of steroid hormone levels were observed both in SLE patients and controls throughout pregnancy, as expected. However, several steroid hormone alterations, in terms of concentration or level profile, were found in SLE patients only, suggesting that endogenous steroid production or metabolism are markedly impaired in SLE during pregnancy and postpartum. Steroid hormone imbalance does not seem to be related to corticosteroid treatment, apart from cortisol levels, which as expected, were significantly lower in patients taking prednisone at a mean dosage >5 mg/day compared with controls (Figure 4).
Therefore the significant persisting decrease in DHEAS concentration observed in SLE patients suggests that adrenal hypofunction and/or hormone abnormal metabolism in SLE is maintained even during pregnancy.
Estradiol and progesterone showed the most relevant alterations in that both were significantly lower than expected in pregnant women with SLE in the second and even more in the third trimester, periods in which these hormones are predominantly secreted by the placenta (12). Thus, the reduction in their serum concentration might suggest placental damage or dysfunction. In fact, placental changes or insufficiency related to a decidual vasculopathy/coagulopathy and/or chronic villitis of unknown etiology are frequently reported in SLE pregnancies, even in the absence of ascertained risk factors, such as antiphospholipid antibodies (39). Alternatively, the findings could be due to a placental metabolism abnormality.
Generally, the most relevant immunologic effect of hormonal modifications in normal pregnancy seems to be a steroid-induced Th2 cytokine polarization (10), i.e., a progressive inhibition of cellular immunity with the consequent enhancement of humoral immunity and antibody production(5–7, 10, 40–43). Given that the immune response in SLE is considered to be Th2 cytokine driven (44) because of the overexpression of Th2-type cytokines, in particular IL-10 (45–47), a disease flare in pregnant patients affected by SLE would be expected, particularly in the third trimester. However, several recent prospective studies reported a low percentage of flares in the third trimester of SLE pregnancy (22, 27, 31–34), and our data are consistent with such recent reports showing a disease activity increase in the second trimester and during the postpartum period (Figure 1).
Therefore, the lower than expected increase in estradiol and progesterone production observed during the third trimester in SLE patients might account for the reduced SLE activity reported in the same gestational period, probably related to the decreased hormone-induced Th2 immune response. In other words, a placental compromise and related reduced hormonal synthesis could be protective against a flare of the disease in SLE pregnant patients, together with the still-evident increase of testosterone, a natural immunosuppressor.
In summary, a dysregulation in steroid hormone production occurs in SLE and becomes more evident during pregnancy. Both gonadal and adrenal hormone levels are reduced during pregnancy in SLE patients compared with healthy controls. The major hormonal alteration observed was an unexpected lack of estrogen (and to a lesser extent progesterone) serum level increase during the second and—even more—the third trimester of gestation, probably due to placental compromise. Therefore, these steroid hormone variations may result in a lower humoral immune activation, probably related to a change in the estrogen/androgen balance, that in turn could account for the decrease in disease activity observed during the third trimester in pregnant SLE patients.