Volume 69, Issue 9 p. 1297-1303
Rheumatoid Arthritis
Free Access

Identifying Clinical Factors Associated With Low Disease Activity and Remission of Rheumatoid Arthritis During Pregnancy

Hilal Ince-Askan

Hilal Ince-Askan

Erasmus Medical Center, Rotterdam, The Netherlands

Search for more papers by this author
Johanna M. W. Hazes

Johanna M. W. Hazes

Erasmus Medical Center, Rotterdam, The Netherlands

Search for more papers by this author
Radboud J. E. M. Dolhain

Corresponding Author

Radboud J. E. M. Dolhain

Erasmus Medical Center, Rotterdam, The Netherlands

Address correspondence to Radboud J. E. M. Dolhain, MD, PhD, Erasmus Medical Center, Department of Rheumatology, Room Nb-852, P.O. Box 2040, 3000 CA Rotterdam, The Netherlands. E-mail: [email protected].Search for more papers by this author
First published: 03 November 2016
Citations: 25

Supported by the Dutch Arthritis Association (Reumafonds).

Dr. Dolhain has received an unrestricted grant from UCB Pharma (more than $10,000).

Abstract

Objective

To identify a combination of clinical factors associated with low disease activity and remission in the third trimester during pregnancy in women with rheumatoid arthritis (RA).

Methods

This study is embedded in the Pregnancy-Induced Amelioration of Rheumatoid Arthritis study, a prospective cohort study. There were data available on 190 pregnancies from first trimester until delivery. Multivariate regression analyses were performed on the disease activity (Disease Activity Score in 28 joints [DAS28] using the C-reactive protein [CRP] level) in the third trimester. Independent covariates were the DAS28-CRP-3 in first trimester, prednisone and sulfasalazine use in the first trimester, parity, methotrexate use in the past, autoantibody status, the presence of erosions, and RA disease duration.

Results

In multivariate regression models, the DAS28-CRP-3, use of prednisone in the first trimester, and the presence of autoantibodies were negatively associated with low disease activity (DAS28-CRP-3 <3.2) in the third trimester (P < 0.05), and the DAS28-CRP-3 and presence of autoantibodies were also associated with remission (DAS28-CRP-3 <2.6) (P < 0.001). Subgroup analysis revealed that the associations of prednisone use and presence of autoantibodies were only present in patients with moderate-to-high disease activity (DAS28-CRP-3 ≥3.2) in the first trimester.

Conclusion

RA patients who have a low DAS28-CRP-3 in the first trimester (irrespective of autoantibody status or prednisone use) are likely to have low disease activity or remission in the third trimester. Also, women with higher disease activity who are not taking prednisone and who express no autoantibodies still have a fair chance of low disease activity in the last trimester.

INTRODUCTION

Pregnancy is the only natural situation that results in spontaneous improvement of rheumatoid arthritis (RA) 1, 2. The most recent prospective studies have shown that RA improves during pregnancy in 40–70% of patients 1, 3-5. However, only 16–27% of patients achieve remission, and approximately 50% still have active disease in the third trimester 1, 4. The treatment of pregnant RA patients is especially challenging. First, achieving and maintaining low disease activity throughout pregnancy is of interest for the welfare of the mother. Second, since several studies have shown that active maternal disease during pregnancy is negatively associated with pregnancy outcome, maintaining low disease activity is even more important 6, 7.

Box 1. Significance & Innovations

  • This is the first study on rheumatoid arthritis (RA) disease activity during pregnancy identifying a combination of clinical factors at the beginning of pregnancy associated with low disease activity and remission in the third trimester.
  • The results of this study show that low disease activity in the first trimester, the absence of autoantibodies, and the absence of prednisone use in the first trimester are independently associated with low disease activity in the third trimester.
  • RA patients who have low disease activity in the first trimester, regardless of their autoantibody status and prednisone use, are likely to have low disease activity and/or remission in the third trimester.
  • Women with higher disease activity who do not use prednisone and express no autoantibodies still have a fair chance of low disease activity in the third trimester.

Unfortunately not all antirheumatic drugs can be continued during pregnancy. Currently, mainly prednisone, sulfasalazine, hydroxychloroquine, and increasingly tumor necrosis factor (TNF) inhibitors 8, 9 are prescribed to pregnant RA patients. Furthermore, the use of a minimal amount of medication during pregnancy is preferred, to prevent drug-related side effects for mother and child. Therefore, and in view of the known improvement of RA during pregnancy, many physicians try to taper or to stop medication when a woman conceives. TNF inhibitors are often discontinued during the first weeks of pregnancy, since most safety data mainly concern first trimester exposure 8, 10. In addition, TNF inhibitors with a high affinity for the fetal fragment crystallizable–receptor are actively transported to the fetus during the second and third trimester of pregnancy, which may result in fetal cord/serum levels equal to or higher than maternal levels 9. Expert opinion therefore advises preferably discontinuing these agents before week 20 of gestation 11.

Moreover, an attempt is often made to stop or to taper the more traditional disease-modifying antirheumatic drugs during pregnancy. The rationale for doing so is that some side effects are mainly related to exposure later during pregnancy, e.g., prednisone has been associated with shorter gestational age 6, gestational hypertension, gestational diabetes, and premature rupture of the membranes 12, side effects that are most prominent during third-trimester exposure. Since lowering medication during pregnancy increases the risk of a flare of disease activity 13, 14, it would be useful to identify those female RA patients in whom medication can be tapered safely in the first trimester of pregnancy.

The aim of the current study was to determine which combination of clinical factors at the beginning of pregnancy was associated with low disease activity (Disease Activity Score in 28 joints [DAS28] using the C-reactive protein [CRP] level <3.2) and remission (DAS28-CRP-3 <2.6) in the third trimester. There is evidence, from univariate analyses, that the presence of autoantibodies 5, the initial DAS28-CRP-3 (4), and the use of prednisone 15 are associated with RA disease activity during pregnancy. However, these factors have not yet been analyzed together in a multivariate model and were never studied in association with other relevant clinical factors. In the current study, we performed multivariate analyses using these known factors and possible other factors, e.g., sulfasalazine use, parity of the mother, methotrexate (MTX) use in the past, the presence of erosions, and the duration of RA.

PATIENTS AND METHODS

Study population

This study is embedded in the Pregnancy-Induced Amelioration of Rheumatoid Arthritis (PARA) study, a nationwide prospective cohort study from The Netherlands, conducted to gain insight into the influence of pregnancy on RA disease activity 4. Patients were eligible for inclusion if they fulfilled the American College of Rheumatology 1987 revised criteria for RA 16 and had a good understanding of the Dutch language. From 2002 to 2008, 475 female RA patients who had a wish to conceive or were already pregnant (in their first trimester) were recruited by their rheumatologist, and from these, 369 were enrolled in the PARA study. During the study period, 205 women conceived at least once.

For the current analysis on identifying associated factors with RA disease activity during pregnancy, patients with a miscarriage (<12 weeks) and patients with a missing DAS28-CRP-3 in the first or third trimester were excluded. After these exclusions, data on 190 pregnancies from 168 women remained available for the current analysis. In 110 pregnancies, a preconception DAS28-CRP-3 was available.

Data collection

In the PARA study, data on disease activity and medication use were collected before conception if possible, each trimester during pregnancy (8–12 weeks, 18–22 weeks, and 28–32 weeks), and 3 times postpartum (at 6, 12, and 26 weeks). The PARA study has been described in detail previously 4. Disease activity was measured with the DAS28-CRP-3 based on the number of swollen joints, the number of tender joints, and the CRP level, without consideration of the visual analog scale of global health 17, 18. Improvement in disease activity was calculated as the difference between first and third trimester DAS28-CRP-3.

Patient characteristics (duration of RA, erosion status, presence of rheumatoid factor [RF], and the presence of anti–citrullinated protein antibody [ACPA]) were obtained from the patient's rheumatologist. RF and ACPA were also determined at study inclusion. A patient was considered to be positive for RF or ACPA when they were positive either at inclusion or in the past 4.

Statistical analysis

For all subjects, descriptive statistics were calculated as numbers, percentages, means, medians, SD scores, and interquartile ranges (IQRs). The mean DAS28-CRP-3 and the mean difference between DAS28-CRP-3 from first to third trimester and from preconception (if available) to third trimester was calculated. Paired t-tests were used to test the differences in disease activity scores.

Two multivariate logistic regression models were built for the analyses of the dichotomized DAS28-CRP-3 in the third trimester (DAS28-CRP-3 ≥ or <2.6 and DAS28-CRP-3 ≥ or <3.2) by step-wise forward selection of covariates with a P value less than 0.2 in the univariate logistic regression analyses. After that, the full models were reduced by eliminating the covariates with a P value greater than or equal to 0.20 in the multivariate regression analyses (if present) to create the final models.

The following independent covariates were analyzed in these models: the DAS28-CRP-3 in the first trimester, prednisone use in the first trimester, sulfasalazine use in the first trimester, parity of the mother, MTX use in the past, autoantibody status (either one or both RF and ACPA positive versus both negative), the presence of erosions, and the duration of RA. The past use of MTX was included in our model as a marker for more severe RA, since in The Netherlands at the time of inclusion for this study, MTX was mainly prescribed to patients with more severe disease. The interaction terms between DAS28-CRP-3 in the first trimester, prednisone use in the first trimester, and the presence of autoantibodies were also analyzed in the final regression models. A stratified analysis was performed if interaction terms were considered significant (P < 0.20). Correlations between the covariates were calculated with Pearson's and Spearman's rank tests.

All of the above mentioned analyses were repeated in patients with a preconception DAS28-CRP-3 available. Instead of the DAS28-CRP-3 in the first trimester and prednisone and sulfasalazine use in the first trimester, the preconception DAS28-CRP-3 and preconception use of prednisone and sulfasalazine were used as independent covariates. P values less than or equal to 0.05 were considered statistically significant, except for interaction terms, where a P value less than 0.20 was considered significant. All statistical analyses were performed using STATA software, version 14.1. This study is in compliance with the Helsinki Declaration, and the Medical Ethics Committee at the Erasmus Medical Center approved the PARA study.

RESULTS

Participants

In total, data on 190 pregnancies from 168 women were used for the current analysis. Descriptive statistics of the study population are shown in Table 1. The mean maternal age in the first trimester was 32.2 years and the median duration of RA was 4.8 years. In total, 133 patients (70.0%) were RF positive, and 116 patients (61.1%) were ACPA positive. Of these patients, 143 (75.3%) were classified as either RF or ACPA positive, or positive for both, and 47 (24.7%) as negative for both. Erosions were present in 118 patients (62.1%).

Table 1. Descriptive statistics of study populationa
Baseline characteristics (n = 190) Values
Age of mother, 1st trimester, mean ± SD years 32.2 ± 3.7
Smoking during pregnancy 15 (7.9)
Duration of 1st trimester RA, median (IQR) years 4.8 (2.1–10.0)
RA-associated autoantibodies
RF positive 133 (70.0)
ACPA positive 116 (61.1)
Either RF or ACPA, or both positive 143 (75.3)
Presence of erosions 118 (62.1)
Parity
Nulliparous 101 (53.2)
Multiparous 89 (46.8)
DAS28-CRP-3 1st trimester, mean ± SD 3.6 ± 1.2
1st trimester DAS28-CRP-3b
<2.6 43 (22.6)
≥2.6 to <3.2 34 (17.9)
≥3.2 to ≤5.1 95 (50.0)
>5.1 18 (9.5)
Methotrexate use in the past 116 (61.1)
Biologic agent use in the past 33 (17.4)
Medication use during 1st trimester
Prednisone only 45 (23.7)
Sulfasalazine only 31 (16.3)
Both prednisone and sulfasalazine 24 (12.6)
Hydroxychloroquine (either alone or in combination) 5 (2.6)
No medication during 1st trimester 85 (44.7)
  • a Values are the number (%) unless indicated otherwise. RA = rheumatoid arthritis; IQR = interquartile range; RF = rheumatoid factor; ACPA = anti–citrullinated protein antibody; DAS28-CRP-3 = Disease Activity Score in 28 joints using C-reactive protein levels.
  • b Disease activity groups are defined according to the European League Against Rheumatism criteria.

RA disease activity during pregnancy

The mean ± SD DAS28-CRP-3 changed during pregnancy from 3.6 ± 1.2 in the first trimester to 3.3 ± 1.2 in the third trimester (P < 0.001) to 3.5 ± 1.1 at 12 weeks postpartum (P = 0.01), and finally 3.4 at 26 weeks postpartum (P = 0.03). In the third trimester, 50.0% of the patients had low disease activity (DAS28-CRP-3 <3.2) compared to 40.5% in the first trimester (P = 0.02). The total number of patients in remission (DAS28-CRP-3 <2.6) increased from 43 (22.6%) in the first trimester to 58 (30.5%) in the third trimester (P = 0.03). In 110 pregnancies (57.9%) there was a preconception DAS28-CRP-3 available. In this subgroup, similar results for disease activity (but in this case from preconception to third trimester) as in the whole group were obtained.

Medication use during pregnancy

In this study, the only antirheumatic medications prescribed during pregnancy were prednisone, sulfasalazine, and hydroxychloroquine. Medication use remained overall similar throughout pregnancy. Prednisone, either with or without sulfasalazine and hydroxychloroquine, was used by 80 patients (42.1%) during at least 1 trimester of pregnancy. Seventy patients (36.8%) used prednisone in the first trimester and 58 (30.5%) continued prednisone throughout the whole pregnancy. The median dose of prednisone was 7.5 mg/day (IQR 5.0–10) in every trimester during pregnancy.

Sulfasalazine, either with or without prednisone and hydroxychloroquine, was used by 62 patients (32.6%) during at least 1 trimester of pregnancy. A total of 57 patients (30.0%) used sulfasalazine in the first trimester, and 51 (26.8%) continued sulfasalazine throughout the entire pregnancy. The median dose of sulfasalazine remained 2,000 mg/day (IQR 1,500–2,000) in every trimester during pregnancy.

Hydroxychloroquine, either with or without prednisone and sulfasalazine, was used by 6 patients (3.2%) during at least 1 trimester of pregnancy. Five patients (2.6%) used hydroxychloroquine in the first trimester, and 4 (2.1%) continued hydroxychloroquine throughout the entire pregnancy. The median dose of hydroxychloroquine was 200 mg/day (IQR 200–400) in the first and second trimester, and 300 mg/day (IQR 200–450) in the third trimester of pregnancy.

In the first trimester, 85 patients (44.7%) did not use antirheumatic medication. From these, 74 patients (39.0%) did not use medication throughout the whole pregnancy. Of the 11 patients who started using antirheumatic medication after the first trimester, 6 started using prednisone and 5 started using sulfasalazine. Five of the patients taking prednisone started using it in the second trimester. The mean DAS28-CRP-3 of the patients using prednisone was 4.8 in the first and 4.6 in the third trimester. All patients who received prednisone after the first trimester had a DAS28-CRP-3 in the first, second, and third trimester above 3.2. In total, 116 patients (61.1%) used MTX, and 33 (17.4%) used biologic agents in the past. In all patients, MTX and biologic agents were discontinued prior to conception.

Factors associated with low disease activity and remission in the third trimester

For clinical purposes the DAS28-CRP-3 in the third trimester was dichotomized for easier interpretation. In addition, the DAS28-CRP-3 in the first trimester was dichotomized to <3.2 or ≥3.2. In the multivariate model of low disease activity (DAS28-CRP-3 <3.2) in the third trimester, moderate-to-high disease activity (DAS28-CRP-3 ≥3.2) in the first trimester, the presence of autoantibodies, and the use of prednisone during the first trimester were negatively associated with low disease activity in the third trimester (P < 0.001, 0.033, and 0.024, respectively) (Table 2). In the final model, the interaction term between the DAS28-CRP-3 in the first trimester and prednisone use in the first trimester was significant (P = 0.094).

Table 2. Third trimester factors associated with RA disease activity: multivariate logistic regression models for DAS28-CRP-3 ≥ or <2.6 and ≥ or <3.2 in the third trimester (n = 190)a
Low disease activity (DAS28-CRP-3 <3.2)b Remission (DAS28-CRP-3 <2.6)b

Univariate

OR

P

Multivariate

ORc

P

Univariate

OR

P

Multivariate

ORc

P
DAS28-CRP-3 1st trimester ≥3.2 0.178d < 0.001d 0.193d < 0.001d 0.092d < 0.001d 0.095d < 0.001d
Sulfasalazine use 1st trimester 1.162 0.635 1.351 0.372
Prednisone use 1st trimester 0.437d 0.007d 0.463d 0.024d 0.488d 0.039d 0.494 0.089
Parity ≥1 1.043 0.884 1.326 0.372
MTX use ever 0.766 0.372 0.779 0.437
Autoantibody status positive 0.369d 0.005d 0.442d 0.033d 0.299d 0.001d 0.316d 0.005d
Presence of erosions 0.638 0.136 0.899 0.740
Duration of RA, years 1.008 0.712 1.002 0.934
Interaction terms
DAS28-CRP-3 1st trimester × prednisone use 1st trimester 0.289d 0.094d
DAS28-CRP-3 1st trimester × autoantibody status positive 1.303 0.743 0.172d 0.033d
Prednisone use 1st trimester × autoantibody status positive 0.674 0.631
  • a RA = rheumatoid arthritis; DAS28-CRP-3 = Disease Activity Score in 28 joints using C-reactive protein levels; OR = odds ratio; MTX = methotrexate.
  • b Based on logistic regression.
  • c Multivariate model after forward selection (limit for inclusion: P < 0.20) and backward selection (limit for exclusion: P ≥ 0.20).
  • d Statistically significant.

In the multivariate logistic regression analysis of remission (DAS28-CRP-3 <2.6) in the third trimester, only moderate-to-high disease activity (DAS28-CRP-3 ≥3.2) in the first trimester and the presence of autoantibodies were negatively associated with remission (DAS28-CRP-3 <2.6) in the third trimester (P < 0.001 and P < 0.005, respectively) (Table 2). Prednisone did not reach statistical significance (P = 0.089), but there was a trend for a negative association. In the final model, the interaction term between the DAS28-CRP-3 in the first trimester and the presence of autoantibodies was significant (P = 0.033). In the subgroup analyses of patients for whom a preconception DAS28-CRP-3 was available (n = 110), similar results as in the whole group were obtained with the preconception variables (data not shown).

Since 2 of the interaction terms were positive in these final models, subgroups based on the DAS28-CRP-3 (<3.2 or ≥3.2) in the first trimester were created to perform a stratified analysis (Table 3). In patients with a low initial DAS28-CRP-3 (<3.2), neither prednisone use in the first trimester nor the presence of autoantibodies were associated with low disease activity or remission in the third trimester. In patients with a DAS28-CRP-3 ≥3.2 in the first trimester, prednisone use in the first trimester was negatively associated with low disease activity, and the absence of autoantibodies was associated with remission in the third trimester in the multivariate models (P = 0.007 and P = 0.001, respectively) (Table 3).

Table 3. Subgroup analysis in patients with DAS28-CRP-3 in the first trimester <3.2 or ≥3.2 with factors associated with RA disease activity in the third trimester, using multivariate logistic regression modelsa
Prednisone use 1st trimester Positive autoantibody status
DAS28-CRP-3 from 1st to 3rd trimester Univariate P Multivariate P Univariate P Multivariate P
<3.2 (n = 77) to low disease activity (n = 57)b
OR 1.077 0.896 0.969 0.957 0.429 0.174 0.427 0.175
<3.2 (n = 77) to remission (n = 45)c
OR 0.606 0.314 0.570 0.266 0.709 0.493 0.654 0.407
≥3.2 (n = 113) to low disease activity (n = 38)b
Coefficient 0.260d 0.003d 0.282d 0.007d 0.422 0.075 0.523 0.198
≥3.2 (n = 113) to remission (n = 13)c
Coefficient 0.398 0.181 0.580 0.460 0.102d < 0.001d 0.112d 0.001d
  • a In addition, all other covariates, which were not significant in the multivariate logistic regression models in Table 2, were tested in these subgroups. However, these covariates were all not significant in these analyses as well (data not shown). DAS28-CRP-3 = Disease Activity Score in 28 joints using C-reactive protein levels; RA = rheumatoid arthritis; OR = odds ratio.
  • b Low disease activity was DAS28-CRP-3 <3.2 in the third trimester.
  • c Remission was DAS28-CRP-3 <2.6 in the third trimester.
  • d Statistically significant.

Based on the results of the stratified analysis, patients were divided into groups to get more insight in the chance of low disease activity and remission in the third trimester (Table 4). Since prednisone use in the first trimester was only associated with low disease activity and the presence of autoantibodies was only associated with remission in patients with an initial DAS28-CRP-3 ≥3.2, only these subgroups were created.

Table 4. DAS28-CRP-3 <3.2 or ≥3.2 in the first trimester: subgroups based on stratified multivariate logistic regression models for low disease activity and remission in the third trimestera
1st trimester prednisone 1st trimester autoantibodies

<3.2

(n = 77)

≥3.2, yes

(n = 46)

≥3.2, no

(n = 67)

<3.2

(n = 77)

≥3.2, positive

(n = 91)

≥3.2, negative

(n = 22)

3rd trimester
Low disease activity: <3.2 57 (74.0) 8 (17.4) 30 (44.8)
Remission: <2.6 45 (58.4) 5 (5.5) 8 (36.4)
  • a Values are the number (%). DAS28-CRP-3 = Disease Activity Score in 28 joints using C-reactive protein levels.

In total, 95 patients achieved low disease activity (DAS28-CRP-3 <3.2) in the third trimester (Table 4). From the 77 patients with a DAS28-CRP-3 <3.2 in the first trimester, the majority (74.0%) had low disease activity in the third trimester as well. From the patients with an initial DAS28-CRP-3 ≥3.2, the chance for low disease activity was the highest in the subgroup which did not use prednisone in the first trimester (44.8%).

In total, 58 patients achieved remission (DAS28-CRP-3 <2.6) in the third trimester (Table 4). From the 77 patients with a DAS28-CRP-3 <3.2 in the first trimester, 45 (58.4%) were in remission in the third trimester. From the 91 patients who had a first trimester DAS28-CRP-3 ≥3.2 and autoantibodies present, only 5 (5.5%) achieved remission. In total 8 of 22 patients (36.4%) with an initial DAS28-CRP-3 ≥3.2 but without autoantibodies achieved remission in the third trimester.

DISCUSSION

Almost all studies conducted so far on the subject of pregnancy in patients with RA have only focused on the occurrence of improvement rather than determining patient characteristics associated with disease activity. In this prospective study, we showed that a low initial DAS28-CRP-3, and the absence of prednisone use and the absence of autoantibodies were associated with low RA disease activity in the third trimester. Other factors we thought to be associated with the disease course that were analyzed in this study (sulfasalazine use in the first trimester, parity of the mother, MTX use in the past, the presence of erosions, and the duration of RA) did not reach statistical significance. Of course there might be other, maybe genetic factors of mother and child that influence the improvement of RA during pregnancy 3, 19, 20. However, this factor would not help us in the daily practice, since we do not have the genetic profile of every patient and her offspring. Additionally, research has recently shown that the disease course in a first pregnancy is not predictive for a subsequent pregnancy 21. We considered the sex of the child as an influential factor, but this factor also did not reach statistical significance (data not shown). On the other hand, the sex is not known until approximately midgestation, and therefore is not useful in this context.

The mean decrease in DAS28-CRP-3 of 0.3 that we found in our study is low, but the reason for this decrease is probably, as previously reported by de Man et al 4, due to the fact that a considerable number of patients already had low disease activity (40.5%) or were in remission (22.6%) in the first trimester. The decrease in DAS28-CRP-3 was the highest in patients with higher disease activity in the first trimester 4, but these patients are still less likely, as shown in this study, to achieve low disease activity or remission.

It could be that prednisone was a proxy for disease activity, since the mean DAS28-CRP-3 was the highest in patients who used prednisone during the first trimester (3.9 versus 3.4). However, of the 18 patients who had high disease activity in the first trimester (DAS28-CRP-3 >5.1), 10 used prednisone and 8 did not use it in the first trimester. From the 10 patients who used prednisone, in total 8 still had high disease activity in the third trimester. From the 8 patients with high disease activity who did not use prednisone in the first trimester, 4 still had high disease activity in the third trimester. Since these numbers are low, we cannot draw any firm conclusions. However, although speculative, we cannot rule out the idea that prednisone itself may inhibit mechanisms that lead to improvement of RA during pregnancy.

Many patients and physicians alike expect to be able to taper medication at the beginning of pregnancy in RA patients, since they rely on the spontaneous improvement of RA during pregnancy. Based on the results of our study, one may consider tapering medication in those patients who are already in remission or have low disease activity at the beginning of pregnancy, and maybe even in those with higher disease activity but not expressing autoantibodies and not taking prednisone. For those patients who still have moderate-to-high disease activity and are taking prednisone or express autoantibodies, we advise continuing medication, if possible, throughout pregnancy. Sound evidence-based recommendations on tapering medication can only be given based on a controlled trial in which medication is tapered in a randomized manner. Given the restrictions of pregnancy, such a trial is very unlikely to be conducted. If such a trial does take place, the patient characteristics identified in this study could be used for patient stratification.

Finally, our study has some limitations. First, none of the patients within this cohort used biologic agents during pregnancy. It would be interesting to replicate the results of our current study in a cohort with patients who used biologic agents during at least the first trimester of pregnancy. Second, in 42% of the patients, there was no preconception DAS28-CRP-3 available. Nevertheless, the regression analysis in the subgroup of 110 patients with a preconception DAS28-CRP-3 reached comparable statistically significant results to the whole group. Furthermore, there was no significant difference between preconception DAS28-CRP-3 and the first trimester DAS28-CRP-3 in these patients (P = 0.43). Third, since rheumatologists in this study were free to change medication and dosages, lowering or discontinuing prednisone after the first trimester could cause an increased DAS28-CRP-3 in the third trimester, thereby resulting in an association between prednisone use during the first trimester and more active disease during the third trimester. However, our analysis showed that this association was not the case in our patient group. In total, 12 patients who used prednisone in the first trimester did not use prednisone in the third trimester. From these, only 1 patient had low disease activity (DAS28-CRP-3 <3.2) in the first trimester and ended with a higher DAS28-CRP in the third trimester (DAS28-CRP-3 = 4.48).

Furthermore, in total 14 patients used a lower prednisone dose in the third trimester compared to the first trimester. From these, only 2 had low disease activity (DAS28-CRP-3 <3.2) in the first trimester and ended with a higher DAS28-CRP-3 in the third trimester (DAS28-CRP-3 3.46 and 3.70, respectively). These findings are not likely to have induced bias in the analyses, and therefore we assume that the negative association between prednisone use in the first trimester and the DAS28-CRP-3 in the third trimester is plausible. Fourth, dichotomizing the third trimester DAS28-CRP-3 to ≥ or <2.6 and ≥ or <3.2 resulted in slight loss of power in the association between prednisone use and remission, but there was a trend for significance (P = 0.08). Finally, for the current study we used all 190 pregnancies from 168 women to enlarge the power. To exclude possible selection bias in the obtained results, we performed a subgroup analysis, including only the first participation of the patients (n = 168), and found similar results in all regression analyses (data not shown).

In conclusion, our study shows that a low initial DAS28-CRP-3 (preconception or first trimester), the absence of autoantibodies, and the absence of prednisone use at preconception or in the first trimester are associated with low disease activity in the third trimester. These findings create more insight into the phenomenon of spontaneous improvement of RA during pregnancy and can be helpful in the daily care for our pregnant RA patients.

ACKNOWLEDGMENTS

The authors thank all patients for participating in the PARA study, as well as the Dutch rheumatologists for their contributions to the study. The authors also thank Yaël de Man, Fleur van de Geijn, and all research assistants and laboratory workers for their help with data collection.

    AUTHOR CONTRIBUTIONS

    All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Dolhain 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 conception and design

    Ince-Askan, Hazes, Dolhain.

    Acquisition of data

    Hazes, Dolhain.

    Analysis and interpretation of data

    Ince-Askan, Dolhain.