Article Text

Download PDFPDF

Extended report
Circulating maternal cytokines influence fetal growth in pregnant women with rheumatoid arthritis
  1. Florentien D O de Steenwinkel1,
  2. Anita C S Hokken-Koelega2,
  3. Yaël A de Man1,
  4. Y B de Rijke3,
  5. Maria A J de Ridder4,
  6. Johanna M W Hazes1,
  7. Radboud J E M Dolhain1
  1. 1Department of Rheumatology, Erasmus MC, Rotterdam, The Netherlands
  2. 2Department of Pediatrics, Subdivision of Endocrinology, Erasmus MC/Sophia Children's Hospital, Rotterdam, The Netherlands
  3. 3Department of Clinical Chemistry, Erasmus MC/Sophia Children's Hospital, Rotterdam, The Netherlands
  4. 4Department of Biostatistics, Erasmus MC, Rotterdam, The Netherlands
  1. Correspondence to Florentien de Steenwinkel, Department of Rheumatology, Erasmus MC, PO Box 2040, Room Ee 963a, Rotterdam 3000 CA, The Netherlands; f.desteenwinkel{at}


Background High rheumatoid arthritis (RA) disease activity during pregnancy is associated with a lower birth weight. Active RA is characterised by high circulating levels of cytokines, which can mediate placental growth and remodelling.

Objectives To assess the influence of maternal serum cytokine levels on birth weight in RA pregnancy.

Methods This study is embedded in the PARA Study, a prospective study on RA and pregnancy. In the present study, 161 pregnant women with RA and 32 healthy pregnant women were studied. The main outcome measures were birth weight SD score (birth weight SDS) in relation to maternal serum levels of interleukin-10 (IL-10), interleukin-6 (IL-6) and tumour necrosis factor-α (TNFα) at three different time points: preconception and during the first and third trimester. Single-nucleotide polymorphisms (SNPs) in the corresponding cytokine genes were also studied.

Results During the first trimester, IL-10 was detectable in 16% of patients with RA, IL-6 in 71%, and TNFα in all patients with RA. Mean birth weight SDS of children born to mothers with RA was higher when IL-10 level was high compared with low (difference=0.75; p=0.04), and lower when IL-6 was high compared with low (difference=0.50; p<0.01) in the first trimester. No correlation was seen at the other time points studied or with TNFα. Cytokine levels were not related to their corresponding SNPs.

Conclusions Maternal IL-10 and IL-6 levels are associated with fetal growth in RA. In the first trimester, high IL-10 levels are associated with higher birth weight SDS, and high IL-6 levels are associated with lower birth weight SDS, even after correction for disease activity.

  • Rheumatoid Arthritis
  • Cytokines
  • DAS28
View Full Text

Statistics from


In normal pregnancies, cytokine levels are not always detectable, but active rheumatoid arthritis (RA) is characterised by high serum levels of cytokines. RA is a chronic, systemic autoimmune disease in which cytokines play a crucial role in the disease activity.1

High maternal cytokines can initiate and intensify the cascade of inflammatory cytokine production during normal pregnancy, resulting in maldevelopment of the placenta leading to spontaneous abortion, intrauterine growth restriction (IUGR) or preterm delivery.2 Aberrant maternal cytokine levels have been reported in several perinatal complications, such as pre-eclampsia,3 preterm delivery4 and threatened miscarriages.5

It has been shown that high RA disease activity is associated with a lower birth weight.6 High disease activity can be related to high cytokine levels, but it is not known if high cytokine levels influence birth weight. A lower birth weight, even within the normal range, might be a risk factor for perinatal complications and is associated with developmental delay, cardiovascular disease, hypertension, non-insulin-dependent diabetes mellitus and neuropsychiatric disorder in adulthood.7–9

In the present study, three different points in time were chosen to evaluate cytokine levels: preconception, first and third trimester. If high circulating cytokines influence placentation and even implantation of the fetus, as mice studies have emphasised,10 the focus should be on the beginning of pregnancy. The third trimester was chosen because of the rapid growth and maturation of the fetus in that period. Interleukin-10 (IL-10), interleukin-6 (IL-6) and tumour necrosis factor-α (TNFα) were studied, because they are implicated in the pathogenesis of RA as well as in pregnancy outcome.1 ,5 ,11–13

Certain functional variations due to single-nucleotide polymorphisms (SNPs) in the genes of IL-10, IL-6 and TNFα have been shown to influence cytokine levels, increase the risk of developing RA, and be associated with preterm birth and increased risk of pre-eclampsia.14 In our study, these specific SNPs were taken into account.

We hypothesised that high maternal serum levels of cytokines influence fetal growth during pregnancy in women with RA. Describing these effects should lead to a better understanding of the function of cytokines during pregnancy and their influence on pregnancy outcome.


Study population

This study is embedded in the PARA Study (Pregnancy-induced Amelioration of RA), a prospective, nationwide cohort study in pregnant women with RA.15 For the present analyses, information was obtained at preconception (range of 3 month–1 year before conception) and during the first trimester (range 8–12 weeks of gestation) and the third trimester (range 28–32 weeks of gestation). Preconception was defined as actively trying to become pregnant after stopping all teratogenic RA medication for at least 3 months. Women were recruited by their rheumatologist and were eligible for inclusion if they wished to become pregnant or when they were already pregnant, but no further than the second trimester. All women met the American College of Rheumatology 1987 revised criteria for RA.16 Only singleton pregnancies of Caucasian women, who delivered a child without congenital deformities, were included.17 A total of 162 participants were enrolled, of whom 73 were seen at preconception, 139 were visited in the first trimester, and data for 158 were available for the third trimester. One patient was excluded from the study because of having breast cancer during her pregnancy. Therefore, 161 Caucasian women with RA were eligible for the study.

Since serum cytokine levels may depend on the detection method, a comparison group was included to observe the main outcome in a normal population. Thirty-two healthy pregnant Caucasian women without an adverse obstetric history were recruited. They were visited at the same time points and had the same assessments and laboratory tests as the pregnant women with RA. The PARA Study was approved by the Medical Ethics Committee, Erasmus MC (Rotterdam, The Netherlands).

Pregnancy outcome

Data on pregnancy outcome included birth weight, gestational age at delivery, and gender of the child. Birth weight was expressed as birth weight SD scores (birth weight SDS), corrected for gestational age and gender (see online supplementary figure S1).18

Factors associated with pregnancy outcome

Clinical characteristics were collected by medical record and physical examination. They included maternal age, gynaecological history, medication use and smoking habits.

At every time point, RA Disease Activity Score (DAS28) was calculated by examining 28 joints using three variables: number of swollen joints, number of tender joints, and serum C-reactive protein (CRP) level.19 It has been shown that RA disease activity is most reliably assessed with this modality of the DAS28 during pregnancy.20 CRP levels were directly measured using the Tina-Quant CRP Immunological Test System (Roche Diagnostics, Almere, The Netherlands). After centrifugation, all samples were frozen at −80°C until assayed. Serum levels of IL-10, IL-6 and TNFα were determined using the immunoassay system, IMMULITE 1000 (Siemens Healthcare Diagnostics, Breda, The Netherlands). The intra-run/inter-run mean±variation coefficient was: IL-10, 27.7±4.6%/23.35±5.4%; IL-6, 105.6±4.9%/87.67±6.1%; TNFα, 105.6±4.9%/89.5±3.8%. The lower limit of quantification was IL-10, 5.0; IL-6, 2.0; TNFα, 2.0. All cytokine levels are presented in pg/ml.

Genetic factors associated with cytokine levels

SNPs selected were either proven to be functional in relation to cytokine level or associated with RA or pregnancy outcome. In addition, a minor allele frequency of >0.10 in the National Center for Biotechnology Information database was required.21 Eight SNPs were selected, which all proved to be located in the promoter region of the cytokine gene. Blood for DNA isolation was available for 134 patients.

On chromosome 1, the IL10 gene has three SNPs known to influence serum IL-10 levels: rs1800871, rs1800872 and rs180089614 The last of these SNPs increases the risk of developing RA when it is a homozygous AA carrier.22

On chromosome 7, the IL6 gene has two SNPs, rs1800795 and rs1800797, which are known to influence serum IL-6 levels and are associated with risk of preterm birth.23

On chromosome 6, the TNF gene has three SNPs known to influence serum TNFα levels: rs1800630, rs1800629 and rs1799724.24 To evaluate rs1800630, we took a tagging SNP in high linkage disequilibrium (rs2844482; r2=1.00).25 All genetic testing was performed using a Sequenom iPLEX.

Statistical analysis

Descriptive statistics are presented as numbers, percentages, means and SDs. Spearman rank correlation coefficients were calculated to evaluate correlations between disease activity and serum levels of cytokines.

Univariate linear regression analyses were performed, with birth weight SDS as dependent variable and maternal serum levels of IL-10, IL-6 and TNFα at preconception and in the first and last trimester as independent variables. Even after transformation, cytokine levels were not normally distributed and were therefore dichotomised according to their median: 0 when the level was lower or equal to the median (low), and 1 when the level was higher than the median (high). When the median was 0, the median of the detectable cytokine levels was taken. Depending on the trimester, IL-10 was measurable in 11–16% of the women with RA (table 1). Consequently, the median of the measurable levels was taken. For IL-6 and TNFα, the median of all levels was taken. To describe the association between maternal cytokine level and birth weight SDS, RA disease activity was addressed as a confounder because it is known to influence birth weight and cytokine levels.26 Other potential confounders are maternal age, smoking habit, and use of prednisone and/or sulfasalazine.27 A multivariate regression model (analysis of covariance) with backward selection was used, and all potential confounders with highest p value were eliminated from the model until all p values were less than 0.2.

Table 1

Patient and pregnancy characteristics

For SNP genotype distributions, significant departure from Hardy–Weinberg equilibrium was calculated using the χ2 test. The Lewontin's D prime (D′) and correlation coefficient (r²) were calculated to assess the presence of linkage disequilibrium (r²≥0.8). The Kruskal–Wallis rank test was used to determine the difference in serum cytokine levels within the three genotype groups. Linear regression, based on allele dose, was performed, with birth weight SDS as dependent variable and genotype groups as independent variable. All statistical analyses were performed using Stata software (V.12.0 for Windows). All SNP analyses were performed using Haploview (V.4.2 for Windows) with the CEU trio as reference.



Mean maternal age at delivery was 32.5 years, and mean RA disease duration was 7.9 years (table 1). Fewer than 4% of the subjects smoked, and 52–55% used medication during pregnancy. Medication use was restricted to prednisone, sulfasalazine or both (table 1).

None of the pregnant women used methotrexate or biological agents before conception or during pregnancy. Hydroxychloroquine was only used by a minority of patients (n=3).

Mean birth weight SDS was −0.004 SD (1.09). As expected, higher DAS28 was associated with a lower birth weight.6 During the third trimester, an increase of 1 point in DAS28 correlated with a decrease in birth weight SDS of 0.21 (p=0.005).

The 32 healthy pregnant women had a mean age of 32.1 years, and 10% (3/32) smoked. Mean birth weight SDS was −0.02 SD (0.96) (table 1).

Cytokine levels and birth weight

All cytokine levels correlated strongly with each other, and the levels at different time points were related (table 2). Levels of IL-10 and IL-6, but not TNFα, correlated with DAS28 during all time points (first trimester shown in table 2). Cytokine levels in healthy women showed similar correlations, but these were not significant, because of the small sample (data not shown).

Table 2

Correlation of cytokine levels, DAS in the first trimester and other time point in patients with RA

As stated above, cytokine levels were defined as low when the level was lower or equal to the median, and high when the level was higher than the median. When the median was 0, the median of the detectable cytokine level was taken.

During the first trimester, IL-10 was detectable in 16% of patients with RA (n=21). According to our definition, IL-10 was high in 7% of the patients (n=10) and low in 93% (n=129). The mean birth weight SDS was 0.55 in the high-IL-10 group and −0.07 in the low-IL-10 group. The high-IL-10 group was associated with higher birth weight SDS than the low-IL-10 group (difference 0.75, p=0.040) after adjustment for confounders (table 3). No such effect was seen at preconception or in the third trimester.

Table 3

Effect of dichotomised (high vs low) cytokine levels (crude and adjusted) on birth weight SDS

During the first trimester, IL-6 was detectable in 71% of the patients with RA (n=95). It was high in 48% (n=67). The mean birth weight SDS was −0.26 in the high-IL-6 group and 0.21 in the low-IL-6 group. The high-IL-6 group was associated with lower birth weight SDS (difference −0.50, p=0.009) after adjustment for confounders (table 3). No such effect was seen at the other time points (table 3). When the effects of IL-6 and IL-10 were analysed simultaneously, the effect was even more pronounced, resulting in a birth weight SDS increase of 0.82 for high IL-10 and a decrease of 0.58 for high IL-6 (table 4).

Table 4

Analysis of covariance: effect of dichotomised (high vs low) cytokine levels in the first trimester on birth weight SDS (levels put in the model simultaneously)

High levels of TNFα had no effect on birth weight SDS at any time point (table 3). Cytokine levels did not influence gestational age (data not shown). In the reference group, cytokine levels were low and often undetectable (table 1). Numbers were too small to allow statistical analysis.

Single-nucleotide polymorphisms

Genotypes across all eight SNPs had no effect on maternal cytokine levels even after correction for disease activity or medication use. Two SNPs were in Hardy–Weinberg disequilibrium, and two had linkage disequilibrium (table 5). Mothers who were homozygous for A in the polymorphism rs1800896 had newborns with a significantly (p=0.021) higher birth weight SDS (0.27 (SD±1.02)) compared with mothers homozygous for G (−0.34 (SD±1.03)). Mean birth weight SDS in the heterozygous group was −0.14 (SD±1.11).

Table 5

Effect of genotype on cytokine levels and birth weight SDS


We hypothesised that high maternal serum cytokine levels influence fetal growth in pregnant women with RA. This hypothesis appears to be true. In our prospective study, we observed that both IL-10 and IL-6 levels influence fetal growth in pregnant women with RA. During the first trimester, high IL-10 levels are associated with higher birth weight SDS, and high IL-6 levels are associated with lower birth weight SDS. This association was still present after correction for disease activity, maternal age, smoking, and use of prednisone and/or sulfasalazine during pregnancy. Unlike most studies, our study focused not only on the last trimester of the pregnancy, but also preconception and the first trimester, emphasising the importance of cytokine levels at the beginning of pregnancy.

Birth weight SDS was 0.75 higher in the high-IL-10 group than in the low-IL-10 group. Birth weight SDS was 0.50 lower in the high-IL-6 group than the low-IL-6 group. The effect of high levels became even more prominent when IL-10 and IL-6 levels were analysed simultaneously. Birth weight SDS was 0.82 higher in the high-IL-10 group and 0.58 lower in the high-IL-6 group. A difference of 0.5 SDS is considered to be of clinical relevance, and therefore our findings are of relevance in women with high IL-10 and high IL-6 at the beginning of their pregnancy.

In our study, cytokine levels were independent of their functional SNPs. We explored the possibility of predicting birth weight based on preconception cytokine levels or functional variations in the relevant cytokine genes. However, no relation was found. These gene factors therefore describe insufficient variance to be useful in prediction. In addition, we observed that only the IL10 gene polymorphism, rs1800896—which in other studies is related to higher IL-10 levels14 ,22—was associated with birth weight, but it cannot be excluded that this is attributable to multiple testing

To evaluate the effects of high cytokine levels, we studied an RA population that is often characterised by high serum cytokine levels. In our RA population, 16% had a detectable IL-10 level comparable with literature values.28 Even though the literature indicates that different diseases may be associated with specific cytokine profiles during pregnancy,29 the results of this study may still produce a better understanding of the consequences of high cytokine levels during pregnancy in general, but particularly in the first trimester.

A comparison group of 32 healthy pregnant women were included in the study. The main purpose of this reference group was to observe maternal serum cytokine levels in normal pregnancies. The number is too small to draw conclusions.

This study highlights the influence of cytokines throughout the first trimester of pregnancy, during which placentation occurs. The role of interleukins has not yet been fully clarified, but we hypothesise that cytokines influence placentation in various ways, by directly influencing placental growth factor (PlGF) and vascular endothelial growth factor (VEGF) and by indirectly influencing extracellular matrix (ECM) degradation. During normal placentation, cytotrophoblasts from the fetal stem cells differentiate and invade the uterine wall by destroying and displacing the ECM, which anchors the placenta to the uterus and provides maternal blood to the fetus. The presence and function of PlGF and VEGF are critical during placentation.30 Anti-inflammatory cytokines such as IL-10 are considered to have a PlGF-enhancing effect.31 Proinflammatory cytokines such as TNFα and IL-6 have adverse effects on cytotrophoblast growth and development.31

One of the main mediators of ECM degradation is metalloproteinase-9 (MMP-9)32 Overexpression of MMP-9 is suggested to contribute to ECM degradation in the fetal membrane and placenta, thereby reducing placental function and even causing fetal membrane rupture or placental detachment.33 Previous research has shown that IL-10 inhibits the expression of MMP-9 in cytotrophoblasts.34 Other studies have shown elevated levels of MMP-9 in pregnant women with RA.35 In our population, maternal IL-10 levels were more often detectable because of their RA. High IL-10 levels positively influence fetal growth, and therefore our data may support the idea that IL-10 participates in the regulation of trophoblasts during human pregnancy by inhibiting MMP-9, thereby creating a more vital placenta. It also supports the idea that the MMP-9 content in first-trimester trophoblasts is more important than in third-trimester cells.34

High RA disease activity is associated with lower birth weight.6 There is a theory that high IL-10 levels are responsible for the improvement in RA during pregnancy.28 One could speculate that the effect of IL-10 levels on birth weight is not related to their effect on placentation, but acts indirectly by reducing maternal disease activity during pregnancy. However, in our cohort, high IL-10 levels were not associated with an improvement in RA during pregnancy; patients with high IL-10 had higher DAS28 during both in the first and third trimester than those with low IL-10.

Our main findings on IL-10 levels are consistent with mice studies. IL10 knockout mice were found to have more adverse pregnancy outcomes.36 Fetal loss and fetal growth restriction have been shown to be attenuated by the administration of IL-10 to mice with endotoxin-induced IUGR.10 These workers also showed that serum concentrations of TNFα and IL-6 were considerably higher in IL10 knockout mice than in the wild-type. This suggests that IL-10 modulates resistance to inflammatory stimuli by downregulating expression of proinflammatory cytokines such as TNFα and IL-6, protecting against inflammation-induced pathology.37

Finally, our study has some limitations. First, the biological effects of cytokines are determined by the interplay of cytokines and their soluble receptors. The levels of these soluble receptors were not determined. Second, it is known that body mass index may influence cytokine levels as well as pregnancy outcome and therefore it could be a potential confounder.38 The body mass index of the participants was not recorded in this study.

In conclusion, our study shows that high levels of maternal IL-10 and IL-6 influence the birth weight of babies born to pregnant women with RA. The effect of high IL-10 and IL-6 levels is most prominent in the first trimester when placentation occurs. As high IL-10 and IL-6 levels influence birth weight, they may influence the future development of the child.


We would like to acknowledge all patients, Dutch rheumatologists and obstetricians for their voluntary contributions to the PARA Study. We are grateful to the research assistants, for data collection, and all laboratory technicians, for their assistance with laboratory research. We would also like to thank Professor A G Uitterlinden, for critically reading the manuscript, and Dr F Rivadeneira, for help with analysing genotype data (both from the Department of Internal Medicine, Subdivision of Genetic Laboratory Endocrinology, Erasmus MC, Rotterdam, The Netherlands).


View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Handling editor Tore K Kvien

  • Contributors YdM and RJEMD recruited patients. FDOdS and RJEMD drafted the manuscript and had full access to all of the data in the study. All authors contributed to interpretation of the results.

  • Funding This study was supported by the Dutch Arthritis Association (Reumafonds), a non-profit fund-raising organisation (DAA 08-1-306). The sponsors had no role in the design and conduct of the study, collection, management, analysis, and interpretation of the data, or in the preparation, review, or approval of the manuscript.

  • Competing interests None.

  • Ethics approval Medical Ethics Committee, Erasmus MC (University Medical Centre Rotterdam, Rotterdam, The Netherlands).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.