According to common knowledge and retrospective studies, approximately 75–90% of patients with rheumatoid arthritis (RA) will improve during pregnancy. Prospective data on disease activity during pregnancy are limited. Therefore, this study aimed to prospectively determine the disease activity during pregnancy in RA patients treated in an era of new treatment options.
For 84 RA patients (American College of Rheumatology criteria), a Disease Activity Score in 28 joints (DAS28) and medication use were obtained, before conception if possible, at each trimester of pregnancy and at 6, 12, and 26 weeks postpartum. Improvement and deterioration were determined by assessing changes in DAS28 and by applying the DAS28-derived European League Against Rheumatism (EULAR) response criteria.
Disease activity decreased with statistical significance (P = 0.035) during pregnancy and increased postpartum. In patients with at least moderate disease activity in the first trimester (n = 52), at least 48% had a moderate response during pregnancy according to EULAR-defined response criteria. In patients with low disease activity in the first trimester (n = 32), disease activity was stable during pregnancy. Thirty-nine percent of patients had at least a moderate flare postpartum according to reversed EULAR response criteria. Less medication was used during pregnancy compared with before conception and compared with postpartum.
This study demonstrates that patients achieve remission during pregnancy and deteriorate postpartum, although less frequently than previously described.
Disease activity in patients with rheumatoid arthritis (RA) decreases spontaneously during pregnancy and increases postpartum. These generalities concerning RA rely on previous retrospective studies, which have been largely based on women's self-reports and doctors' observations in an era of restricted treatment options. In the last 7 decades, a substantial number of pregnant patients with RA im proved in those studies; however, this number declined over time from 90% to 53% (1).
Until now, only 1 large, prospective study on this subject has been conducted from late pregnancy onward to 6 months postpartum (2). In retrospect, 63% of patients reported improvement in disease activity. At the third trimester, only 16% were in total remission, which was defined as having no swollen joints and receiving no antirheumatic therapy at the time of measurement. That study had 2 main disadvantages. First, objective information on disease activity was not available before conception or during early pregnancy. Second, no validated scoring systems for disease activity and remission were used. In addition, new treatment options are now available to reduce disease activity before pregnancy, which was not the case with the previous study.
For this reason, we conducted a prospective nationwide cohort study from preconception or early pregnancy onward. This study, the Pregnancy-induced Amelioration of Rheumatoid Arthritis (PARA) study, had 2 main aims. One aim was to measure disease activity in patients with RA using a validated instrument (Disease Activity Score in 28 joints [DAS28]). The second aim was to collect patients' blood and urine samples to perform laboratory investigations into the underlying mechanisms of spontaneous improvement during pregnancy and relapse postpartum.
Here we present the first data of this study regarding the following 3 objectives. The first objective was to calculate the percentages of women in remission during pregnancy and postpartum. The second objective was to prospectively measure whether disease activity decreases during pregnancy and increases postpartum. Disease activity and remission were defined according to the DAS28 with 3 variables: swollen joint count, tender joint count, and C-reactive protein (CRP) level (mg/liter) (DAS28-CRP-3). The third objective was to provide a clear description of what medication was used before, during, and after pregnancy.
PATIENTS AND METHODS
All rheumatologists working in The Netherlands were contacted by mail twice a year from the start of the study in May 2002 until November 2006. Rheumatologists were asked to recruit patients with RA who wanted to conceive or who were already pregnant (preferably in their first trimester). Patients were eligible for the study if they fulfilled the 1987 revised criteria of the American College of Rheumatology (ACR; formerly the American Rheumatism Association) for RA (3) and had a good understanding of the Dutch language. During the study, the patients' own rheumatologists provided patient care. The present analysis used data from patients who had information available from prepregnancy or the first trimester to 26 weeks postpartum onward. Women who had a miscarriage in the first trimester or were pregnant with twins were not included. No woman was included twice.
Patients were visited 6 or 7 times at their home address. They were visited before conception (if possible), at each trimester (8–12 weeks of gestation, 18–22 weeks, and 28–32 weeks), and 3 times postpartum (6 weeks, 12 weeks, and 26 weeks). The visit before conception took place if the woman had a desire to conceive, i.e., she and her partner did not use any contraceptives. If a woman did not conceive within a year after the first visit, another visit took place. Data collected before conception were classed as “before pregnancy.”
Medical and obstetric history were taken at the first visit by interview. Presence of rheumatoid factor (RF), antibodies to cyclic citrullinated peptide (anti-CCP), and erosions were ascertained from the patients' medical records. The presence of anti-CCP, IgM-RF, IgG-RF, and IgA-RF in serum from the first visit was also determined. Presence of anti-CCP was determined with the EliA CCP test (Phadia AB, Uppsala, Sweden) and was defined as a serum level >10 units/ml. Positivity for anti-CCP used in the analyses was defined as “ever recorded positive” in the medical records or “tested positive at first visit.” A positive IgM-RF, IgG-RF, or IgA-RF (Hycor Biomedical, Garden Grove, CA) was defined as a serum level higher than the level at which less than 5% of healthy controls tested positive in these tests. Presence of RF was defined as “ever recorded positive” in the medical records or “tested positive at first visit.”
Current medication use was recorded and percentages of use and median daily doses of medication were calculated per time point. At each visit, women provided information about their current pregnancy, and postpartum they provided information on whether or not their newborn was breastfed.
A research physician or research nurse performed a standardized 28-joint count for swelling and pain. Joint examinations were performed for pain and swelling as recommended by the European League Against Rheumatism (EULAR) Standing Committee for International Clinical Studies Including Therapeutic Trials (ESCISIT) (4). CRP levels were measured in serum by Tina-quant CRP (Roche Diagnostic, Mannheim, Germany).
Disease activity was objectively scored using the DAS28-CRP-3 (5). In a previous study, we demonstrated that disease activity can be measured the most reliably during pregnancy with the DAS28-CRP-3 (6), because erythrocyte sedimentation rate increases physiologically during pregnancy.
Data and statistical analysis.
For all patients, the mean disease activity scores were calculated at the time points indicated. The mean individual difference between DAS28 at the first and third trimesters was calculated. According to the EULAR criteria, remission was defined as a DAS28 <2.6 (7).
Improvement in disease activity between the first and third trimesters was defined, according to the EULAR response criteria, as good, moderate, or no responders as defined by ESCISIT (Appendix A) (6). Because the baseline requirement for the EULAR response criteria is that patients have an initial DAS28 of at least 3.2, 2 subgroups of patients were created, 1 with patients with moderate to high disease activity (DAS28 ≥3.2) and 1 with patients with low disease activity (DAS28 <3.2) in the first trimester. The prerequisite of an initial DAS28 ≥3.2 for application of the EULAR response criteria meant that only those patients with a DAS28 ≥3.2 in the first trimester could be classified as good, moderate, or no responder (4, 7).
Because no classification for deterioration is available yet, we defined flares in disease activity between 6 weeks and 12 or 26 weeks postpartum as severe or moderate based upon “reversed” EULAR response criteria for disease activity (see Appendix A). This classification was applied to all patients; there was no baseline DAS28 requirement. Finally, the mean individual difference between DAS28 at 6 weeks and 12 or 26 weeks postpartum was calculated.
The proportions of women in clinical remission and the proportions with low, moderate, or high disease activity before pregnancy, during pregnancy, and postpartum were calculated. The proportions of women in remission were analyzed by generalized estimating equations with an exchangeable correlation structure. This takes into consideration whether patients are already in remission and stay in remission, or whether other patients go into remission.
All time points during and after pregnancy were used to measure changes in DAS28 over time in a linear mixed model with unstructured covariance. This model considers random variation within individuals and random variation between individuals. In subgroup analyses, the disease course of patients with or without RF, anti-CCP, joint erosions, or biologic agent use prior to their desire to conceive was determined. A likelihood ratio test was used to test the differences in disease course between the different subgroups.
A 2-sided P ≤ 0.05 was considered statistically significant. SPSS for Windows (SPSS, Chicago, IL) version 12.0 was used.
This study is in compliance with the Declaration of Helsinki, and the ethical committee at the Erasmus MC, University Medical Center Rotterdam, Rotterdam, The Netherlands approved this study.
Patient recruitment data.
Up to November 2006, 346 patients were recruited by rheumatologists. Of these, 276 were eligible; however, 111 were not yet pregnant. As shown in Figure 1, we obtained a complete disease activity data set of 84 pregnancies from the first trimester to 26 weeks postpartum. A detailed description of disease activity, medication, and demographics of these patients is provided. Data from 41 pregnancies were available from before pregnancy.
The cohort consisted of patients with RA with a median disease duration of 4.8 years (range 0.1–28.6 years) at first visit. This range can be explained by 11 women who had a desire to conceive within 1 year after diagnosis, and by 7 women who had polyarthritis in childhood that developed into an RF-positive and erosive disease that could be classified as RA. RF was positive in 71% of patients and anti-CCP was positive in 68% (Table 1). Whether or not patients were erosive was documented in 74 pregnancies; of these, 53 (72%) were erosive. Almost half of the pregnancies were first-time pregnancies. Only 3 pregnancies occurred in nonwhite women (2 Asian, 1 African). Postpartum, 67% of the babies were breastfed, 49% for longer than 6 weeks.
Table 1. Clinical and demographic features of 84 rheumatoid arthritis patients and their pregnancies (n = 84)*
Values are the number (percentage) unless otherwise indicated. RF = rheumatoid factor; anti-CCP = anti–cyclic citrullinated peptide antibodies.
Only data for 74 patients were available for this analysis.
Age at delivery, mean ± SD years
31.9 ± 3.3
Disease duration at first visit, median (range) years
The total number of patients in remission (DAS28 <2.6) increased during pregnancy from 17% in the first trimester to 27% in the third trimester (P = 0.16), and decreased to 18% in the 12th week postpartum (P = 0.07) (Figure 2). The number of patients in remission increased, despite the remarkable changes in medication use that took place (Table 2). The percentage of patients who received methotrexate <1 year before conception (28%) was reduced to 0% during pregnancy, and the percentage of patients who received nonsteroidal antiinflammatory drugs (NSAIDs) before conception (32%) was reduced to 4% during pregnancy (data not shown in Table 2). The total number of patients with low disease activity (DAS28 2.6–3.2) remained stable between the first and third trimesters, but decreased postpartum. Almost half of the patients had at least low disease activity in the third trimester (Figure 2). The total number of patients with moderate disease activity (DAS28 3.2–5.1) decreased during pregnancy, while the total number of patients (n = 10) with high disease activity (DAS28 >5.1) remained stable during pregnancy.
Table 2. Disease-modifying antirheumatic drug and biologic agent use before pregnancy, during pregnancy, and postpartum*
Altogether, mean disease activity scores decreased during pregnancy and increased postpartum (n = 84; P = 0.035) (Figure 3). The impact of pregnancy on disease activity in the third trimester was most pronounced in patients who had moderate to high disease activity in the first trimester (mean ± SD −0.54 ± 0.92 [n = 52], DAS28 4.4–3.9; P < 0.001) compared with those with a low disease activity in the first trimester (mean ± SD 0.23 ± 0.92 [n = 32], DAS28 2.5–2.7; P = 0.21). The mean DAS28 increased from 6 weeks to 12 weeks postpartum by 0.30 (95% confidence interval [95% CI] 0.07; 0.48, P = 0.01) from 3.5 to 3.8. In subgroup analyses, the presence of anti-CCP, RF, and erosions did not alter the course of disease activity during pregnancy and postpartum (likelihood ratio tests: P = 0.91, P = 0.98, and P = 0.98, respectively). Biologic agents were used in 6 patients <1 year prior to conception. Because of the small number, no valid statistical analyses could be performed.
The analysis of patients who were visited before pregnancy (n = 41) showed that the mean disease activity before pregnancy decreased in the third trimester by 0.4 (95% CI −0.6, −0.03; P = 0.03) from 3.8 to 3.4. The change in DAS28 from before pregnancy to first trimester was not statistically significant in these patients (mean −0.08 [95% CI −0.3, 0.2]; P = 0.61). The change in DAS28 from before pregnancy to 26 weeks postpartum was not statistically significant (mean −0.22 [95% CI −0.5, 0.1]; P = 0.13).
Changes in disease activity according to DAS28-derived EULAR response criteria
Patients who fulfilled the prerequisite for application of EULAR response criteria, namely, those with a DAS28 >3.2 in the first trimester (n = 52), were classified as good, moderate, and no responders. This analysis showed that 48% of these patients had at least a moderate response during pregnancy (Table 3).
Table 3. Improvement during pregnancy according to the EULAR response criteria, and therefore restricted to patients with moderate disease activity (DAS28 >3.2) in the first trimester (n = 52)*
Classification (EULAR criteria) between first and third trimester
DAS28 >3.2 at first trimester
Values are the number (percentage). EULAR = European League Against Rheumatism; DAS28 = Disease Activity Score in 28 joints.
Of the 6 patients who received biologic agents <1 year before conception, all had moderate to high disease activity in the first trimester. Similar to the total group (n = 52), 50% had at least a moderate response during pregnancy according to EULAR response criteria and 50% had no response.
Postpartum, RA deteriorated according to the “reversed” EULAR response criteria in 39.3% of all 84 patients (Table 4). Of the patients with moderate to high disease activity (DAS28 >3.2) in the first trimester, 2 (3.8%) patients had a severe flare of RA postpartum, 17 (32.7%) patients had at least a moderate flare, and 33 (63.5%) patients had no flare. Of the patients with low disease activity (DAS28 ≤3.2) in the first trimester, 1 (3.1%) patient had a severe flare of RA postpartum, 13 (40.6%) patients had at least a moderate flare, and 18 (56.3%) patients had no flare.
Table 4. Deterioration postpartum according to reversed EULAR response criteria between 6 weeks postpartum and 12 or 26 weeks postpartum, shown for all patients (n = 84)*
Classification (reversed EULAR criteria)
Values are the number (percentage). EULAR = European League Against Rheumatism.
CRP levels during pregnancy.
During pregnancy, mean ± SD CRP levels decreased from 16.2 ± 15.1 mg/liter in the first trimester to 14.2 ± 15.1 mg/liter in the third trimester (not significant). Mean CRP levels increased from 11.3 ± 13.3 mg/liter at 6 weeks postpartum to 14.8 ± 25.1 mg/liter at 12 weeks postpartum, and decreased to 11.5 ± 13.1 mg/liter at 26 weeks postpartum. These differences were not statistically significant.
Evaluation of medication use.
In Table 2, the percentages of disease-modifying antirheumatic drug (DMARD) use are categorized into ever used, used <1 year before conception, before conception, during pregnancy, and postpartum. The medication used during conception is represented by the columns before pregnancy (n = 41) and first trimester (n = 84). The last column shows the percentage of women who were taking the same medication at 26 weeks postpartum as they were taking at <1 year before conception.
Very few patients changed from one drug to another during pregnancy. The median daily dosage of prednisone was 7.5 mg (range 0.5–20 mg) before conception and did not change throughout pregnancy. The median daily dosage of prednisone increased to 10 mg (range 0.5–20 mg) at 6 weeks postpartum and decreased to 7.5 mg (range 0.5– 20 mg) at 12 and 26 weeks postpartum. The median daily dosage of sulfasalazine was 2,000 mg (range 500–4,000 mg) at all time points.
Biologic agents were prescribed before conception in 10 patients. Six patients received biologic agents <1 year before conception; of these, 2 patients restarted biologic agents postpartum. None of the patients received a combination of biologic agents, prednisone, and methotrexate <1 year before conception.
Except for DMARDs and biologic agents, NSAIDs were used during pregnancy in 4% of patients, while up to 33% of patients needed them for pain relief postpartum. Acetaminophen was used in only 5–10% of patients for pain relief throughout pregnancy and postpartum.
The assumption of rheumatologists, based on previous studies, that almost all patients with RA will experience a remission of disease activity during pregnancy does not seem to be valid today. In this nationwide, prospective study, approximately one-quarter of women were in remission in the third trimester, despite the fact that medication use was remarkably reduced during pregnancy compared with before conception. The mean disease activity scores decreased during pregnancy and increased postpartum, and after applying EULAR-defined response criteria, almost half of the women had at least a moderate response during pregnancy and more than one-third had at least a moderate flare postpartum. The postpartum flare may be underestimated because medication use was remarkably increased after delivery.
This is the first study to use the DAS28 as a validated disease activity score to measure disease activity during pregnancy. Therefore, any comparison with previous studies can only be made indirectly. In the present study, 48% of patients who had at least moderate disease activity (DAS28 >3.2) in the first trimester had an improvement in disease activity during pregnancy. Only one large prospective study has been performed for comparison. In that study, performed by Barrett et al, 140 participants were scored for joint pain and swelling in the third trimester and 1 month and 6 months postpartum (2). Improvement in RA during pregnancy, however, was recorded retrospectively by self-report. In that study, 65% of the patients retrospectively reported an improvement in pain and swelling compared with before pregnancy. In addition, a combination of smaller prospective studies (a total of 66 pregnancies) showed improvement in disease activity in approximately two-thirds of pregnancies (8–12). That improvement was measured either by Camp index or by tender and swollen joint scores that were rated between 0 and 3 (8–12). Three of these studies selected patients with moderate to high levels of disease activity prior to pregnancy (9, 10, 12). In our study, changes in disease activity were also measured subjectively by means of a 3-point Likert scale (improved, stable, or deteriorated) between each visit (data not shown). A total of 77% of women reported an improvement at 1 visit, at least, during pregnancy. A difference in outcome between measuring disease activity by DAS28 or patient self-report might also be explained by the fact that in the latter method, patients may also take into consideration their decreased use of medication during pregnancy.
In this study, 27% of patients were in remission in the third trimester according to the DAS28. This is comparable with the percentages of remission shown in previous studies in which other definitions were used. Barrett et al (2), in a manner quite similar to Nelson et al (9), defined remission as using no antirheumatic drugs and having no painful and no swollen joints. In these 2 studies, 16% and 39% of patients, respectively, were in remission in the third trimester. In our study, women had no swollen and no painful joints in the third trimester in 14% of pregnancies, and in only 8% of all pregnancies, no antirheumatic drugs were used.
In the present study, deterioration of RA postpartum was present in at least 39% of all patients with at least a moderate flare according to reversed EULAR response criteria. At least 53% of our patients noted a deterioration on the Likert scale at 12 weeks postpartum compared with 6 weeks postpartum. Barrett et al demonstrated that 62% had more affected joints postpartum, and in ∼70% of their patients, a deterioration in disease activity was self-reported at 1 and 6 months postpartum (2). In addition, Barrett et al showed that the median number of affected joints significantly increased from 8 during pregnancy to 10 joints at 6 months postpartum. At the same time, in 50% of those patients, treatment was increased postpartum. In our study, methotrexate and biologic agents were added to the prednisone or sulfasalazine therapy in 40% and 11% of patients, respectively, for adequate disease control postpartum.
The differences in improvement rates during pregnancy and relapse rates postpartum can therefore be explained by various factors. First, subjective measurements seem to show higher percentages of improvement and deterioration than the objective measurements of disease activity used in this study. Second, in many previous studies, only patients with active disease before pregnancy entered the study. In the present study, all women with RA entered, and a higher percentage of improvement was shown in women with RA with higher disease activity in the first trimester. Finally, as a consequence of better treatment options, the percentages of improvement and deterioration may also be lower today than previously described, because patients may have lower disease activity before pregnancy compared with previous studies and are treated more aggressively postpartum.
Before conception, medication use in patients was remarkably reduced because most DMARDs are contraindicated during pregnancy. Despite this reduction, a decrease in disease activity was still observed, therefore the observed decrease in disease activity during pregnancy must be considered as a consequence of the beneficial effect of pregnancy itself. Another argument that supports the proposal that pregnancy itself has a beneficial effect on disease activity is that treatment of RA during pregnancy was almost completely restricted to sulfasalazine and low-dose prednisone, and it is known that this combination alone is insufficient to induce remission in patients with RA (13).
For daily clinical practice in rheumatology, this study demonstrated that neither presence of RF, as in the study by Barrett et al (2), nor presence of erosions or anti-CCP was associated with improvement during pregnancy or postpartum flare. The levels of RF and anti-CCP in the PARA study were stable during pregnancy until 6 weeks postpartum; only IgM-RF decreased postpartum when treatment was resumed (14). It has previously been suggested that postpartum flare is associated with the cessation of nursing. This assumption, however, could not be tested in the present study because patients have to stop nursing in order to restart treatment when they flare. Furthermore, this study gives rheumatologists additional evidence that patients with low disease activity before pregnancy remain relatively stable, whereas patients with moderate to high disease activity will improve the most.
Up until now, various studies have been undertaken to explain the beneficial effect of pregnancy in RA and many hypotheses have been formulated. Nevertheless, the exact reason is still unknown. Suppression of the immune system by HLA incompatibility between mother and child has been shown to be associated with a favorable disease course (9). In addition, biochemical changes such as elevation of α2-glycoprotein levels during pregnancy (11) and an increase in IgG glycosylation (10) were related to the improvement of disease activity. Finally, as is still hypothesized, a shift in T cell function from a Th1 phenotype to a Th2 may be important (15). The PARA study will provide excellent opportunities for future laboratory studies on autoantibodies and immune regulation during pregnancy and for epidemiologic investigations of pregnancy outcome related to disease activity during pregnancy.
Finally, the PARA study has some limitations. First, the analyses were restricted to women (n = 84) with a complete data set, because medication use could not otherwise be clearly and correctly interpreted. The missing data were, however, completely random, and when the analyses were repeated including all eligible women (n = 100), similar results were obtained. Second, the present analysis was restricted to all data of singleton pregnancies because twin pregnancies result in more physical problems. Third, pregnancy outcome was self-reported; however, this method was verified with medical charts and was found to be valid and accurate for all data reported (16).
In conclusion, this study demonstrates that during pregnancy, patients go into remission and deteriorate postpartum, although less frequently than previously described.
Dr. de Man had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. De Man, Dolhain, Hazes.
Acquisition of data. De Man, Dolhain, van de Geijn, Hazes.
Analysis and interpretation of data. De Man, Dolhain, van de Geijn, Willemsen, Hazes.
Manuscript preparation. De Man, Dolhain, van de Geijn, Willemsen, Hazes.
Statistical analysis. De Man, Dolhain, Willemsen, Hazes.
We acknowledge all patients and Dutch rheumatologists for their voluntary contribution to the PARA study. We are grateful to Esther Gasthuis and Tineke Krommenhoek and other research assistants for their contribution to the data collection. Finally, we would like to thank Professor Th. Stijnen from the Department of Epidemiology and Biostatistics for his advice regarding the statistical analyses.
Table . EUROPEAN LEAGUE AGAINST RHEUMATISM (EULAR) RESPONSE CRITERIA FOR THE DEFINITION OF IMPROVEMENT DURING PREGNANCY (6) AND REVERSED EULAR RESPONSE CRITERIA FOR THE DEFINITION OF DETERIORATION POSTPARTUM
EULAR response criteria
DAS28 at 3rd trimester
Improvement (decrease) in DAS28 from 1st trimester
>0.6 and ≤1.2
>3.2 and ≤5.1
Reversed EULAR response criteria
DAS28 at 12 or 26 weeks postpartum
Deterioration (increase) in DAS28 from 6 weeks postpartum