Abstract
Introduction
Circulating immune markers may be associated with preeclampsia but further investigations in early pregnancy and among preeclampsia subtypes are warranted. We examined immune markers in 208 preeclamptic women and 411 normotensive controls.
Methods
Our study was nested within the Collaborative Perinatal Project. A total of 242 women had first trimester serum samples and 392 had second trimester serum samples. Preeclampsia was defined as hypertension >20 weeks of gestation with proteinuria or pulmonary edema, oliguria, or convulsions. Preterm preeclampsia was defined as preeclampsia with delivery less than 37 weeks of gestation. Associations between immune markers RANTES, interleukin (IL)-6, IL4, IL5, IL12, IL10, IL8, IL1-beta, interferon (IFN)-gamma, tumor necrosis factor (TNF)-alpha and beta, transforming growth factor (TGF)-beta and preeclampsia were explored using a modified version of cox regression developed to address data with non-detectable levels. Models were adjusted for body mass index, gestational age of blood sampling, fetal sex, smoking, socioeconomic status and maternal age.
Results
In first trimester samples, IL-12 was associated with preeclampsia (p=0.0255). IFN-gamma (p=0.0063), IL1-beta (p=0.0006), IL5 (p=0.0422) and TNFr (p=0.0460) were associated with preterm preeclampsia only. In second trimester samples, IL1-beta was associated with preeclampsia (p=0.0180) and term preeclampsia (p=0.0454). After correction for multiple comparisons, only IL1-beta remained associated with preterm preeclampsia in the first trimester (p=0.0288).
Discussion
Elevated first trimester IL1-beta appears to be associated with preterm preeclampsia. However, few associations were observed in the second trimester. Systemic immune markers alone may not be useful for preeclampsia prediction.
Keywords: Inflammation, preeclampsia, pregnancy
INTRODUCTION
Despite progress in understanding preeclampsia pathogenesis, useful prediction models, treatments and specific diagnostic tests have been limited. Preeclampsia is a systemic maternal syndrome that affects 3–10% of pregnancies and is a leading cause of maternal mortality [1]. Clinically, preeclampsia is diagnosed as the new onset of hypertension (≥ 140/90 mm Hg) and proteinuria (≥ 300 mg per 24 hour urine collection) or end-stage organ failure after 20 weeks of gestation and the only treatment is delivery of the placenta [2]. The exact mechanisms that lead to preeclampsia are not completely elucidated. Furthermore, preeclampsia is hypothesized to have several subtypes which can complicate prediction. Thus, identifying biomarkers to better understand pathogenesis and differentiate subtypes will improve clinical management.
In normal pregnancy, the placenta secretes immune stimulating placental derived factors into the maternal circulation [3]. Balanced secretion of these factors may play a role in maternal immune tolerance towards the fetal allograft [4]. However, abnormal placentation could lead to increased placental shedding, exaggerated systemic inflammation and subsequent endothelial dysfunction, the key characteristics of preeclampsia [5, 6]. This is consistent with third trimester studies reporting increased circulating pro-inflammatory markers [tumor necrosis factor alpha (TNFα), interleukin (IL)-8, IL-1β and interferon (IFN)-γ] in preeclampsia [7, 8]. Similar studies have reported associations between pleiotropic or immune-modulatory markers including IL-6, IL-2, and IL-4, as well as anti-inflammatory marker IL10 [7–9]. However, studies conducted prior to the third trimester are conflicting. A second trimester study has reported lower circulating IL10, TNF-α and IFN-γ among women who develop preeclampsia [10]. While others report no significant differences in circulating immune markers [11, 12]. First trimester investigations are limited. Circulating IP-10, a chemokine induced by IFN-γ, is increased in preeclampsia [13]. Additionally, first trimester elevated IL1β was shown to predict early onset preeclampsia [14].
Our own research has shown that mid-trimester systemic immune markers including IL6 and TNFβ are increased in preeclamptic women while IL1β is decreased [15]. This study conducted among 707 women from the Danish National Birth Cohort was unable to fully examine first trimester systemic immune markers. First trimester investigations are important as preeclampsia pathogenesis may begin early in pregnancy. The objective of this paper is to examine the association between first and second trimester circulating immune markers and preeclampsia. Serum immune markers including IL-6, IL-6 receptor, IL-4, IL-4 receptor, IL-5, IL-12, IL-2, TNF-α, TNF-β, TNF-receptor, IL-1β, IL-1α, IL-8, IL-10, IFN-γ, macrophage migration inhibitory factor (MIF), macrophage inflammatory protein (MIP), transforming growth factor-beta (TGF-β), and RANTES were included in our study based on previous associations with preeclampsia or their involvement in systemic inflammation or the Th1/Th17 paradigm [16–18].
METHODS
We conducted a nested case control study of 208 singleton and primiparous preeclamptic women and 411 singleton and primiparous normotensive controls within the Collaborative Perinatal Project (CPP) [19]. Both cases and controls had no history of diabetes, cardiovascular disease or hypertension. The CPP is a longitudinal study of 55,908 pregnancies [20]. Women were enrolled between 1959 and 1965 from 12 university-affiliated medical centers in the United States (Baltimore, MD; Boston, MA; Buffalo, NY; Memphis, TN; Minneapolis, MN; New Orleans, LA; two sites in New York, NY; Philadelphia, PA; Portland, OR; Providence, RI; and Richmond, VA). Oral consent, as was standard at the time of the CPP study, was obtained from all women in the study [21]. We analyzed a total of 242 women who had first serum samples collected in the first trimester (mean gestation age 10.7±1.9; range 5–13 weeks) and 392 women who had first serum samples collected in the second trimester (mean gestation age 16.5±1.7; range 14–19 weeks). The study was approved by the University of Pittsburgh Institutional Review Board.
Data collection
At the first prenatal visit, all women were interviewed in person to obtain data on maternal characteristics, behavior, medical and pregnancy history. Delivery data was recorded by the attending physician. For our analysis, we considered self-reported maternal age (years), marital status (married, single), socioeconomic status, race (white, non-white), maternal smoking (yes/no), and pre-pregnancy body mass index (BMI) as potential covariates. Socioeconomic status was previously determined using a composite score that combined education, occupation and family income [22]. Pre-pregnancy BMI was determined by (weight (kg)/height (m)2 which was reported at enrollment. Gestational week was determined by the date of delivery minus the date of last menstrual period.
Cytokine measurements
Non-fasting blood samples were obtained at the first study visit and stored at −20°C in glass vials and monitored continuously from the time of collection. There were no recorded thaws from collection time until they were aliquoted for analysis. Each sample was labeled with a study ID and name and checked against a pull sheet at the time the samples were pulled. Our subset of samples was mailed to the Statens Serum Institute in Copenhagen where immune biomarkers were measured in duplicate with an in house multiplex flow cytometric assay system Luminex MultiAnalyte Profiling Technology (LabMap, Luminex Corporation, Austin Texas) [23]. The calibration curves for each analyte were calculated by the Bio-Plex 3.0 software (BioRad, US). Mean intra- and interassay CVs (CV %) were 6.2% and 16%, and ranged from 6.7 (IL-4) to 13 (IL-10 and TNF-α) and 10 (IL-4) to 25 (TNF-α) [23]. Variation in precision profiles among analytes is similar to other studies [24–26]. We acknowledge that the long-term storage of CPP samples raises concerns about the measurement of the analytes. This multiplex assay has been demonstrated to be valid for the measurement from specimens obtained from long-term storage using10 anonymously collected residual dried blood spot specimens stored for 23 years at −24°C in a national, Danish biological specimen bank [23]. In that study the measurable amounts of most cytokines were constant. Additionally, analytes measured in serum from the CPP cohort have shown to be stable in other studies [27–29]. In another CPP study, cytokines measured in the CPP were compared to fresh samples and found to be consistent across groups [30]. As IL2 and IL1α measured below the LOD for greater than 75% of patients, these biomarkers were not analyzed.
Preeclampsia definition
Preeclampsia was based on chart abstraction of blood pressure and protein levels and defined as gestational hypertension (2 or more measurements of systolic blood pressure >=140 mmHg and/or diastolic blood pressure >=90 mmHg for the first time after 20 weeks of gestation) and proteinuria (2 random urine dipsticks of 1+ protein or one dipstick of 2+ protein). In the intrapartum period, the first 5 pressures obtained after hospital admission for delivery were averaged. It is accepted that preeclampsia is heterogeneous disease with subtypes (early/late onset) that have different pathophysiological pathways [31]. For our study, we classified preeclampsia resulting in either a term birth (≥37 weeks gestation) or a preterm birth (< 37 weeks gestation) as separate outcomes. Preeclampsia with preterm birth is a valid proxy for disease severity and early onset of disease. Preeclampsia with preterm birth <34 weeks of gestation was not used as the sample size was too small for analysis.
Statistical analyses
For all biomarkers, raw median levels and ranges were calculated for preeclamptic and normotensive women. Our primary analysis examined the association between immune biomarkers and preeclampsia. As circulating immune biomarkers display heterogeneity by gestational age, we stratified our analysis by first and second trimester samples. In multiplex assays, subjects frequently measure outside of the limit of detection (LOD) [32]. In general, the frequency of women who measure beyond the LOD varies and those who are outside the detectable range may have very low or undetectable levels. Common approaches to handling data below the LOD such as replacing with a set value (0 or LOD/2), dichotomizing (above LOD vs. below LOD) or multiple imputation are well recognized to be prone to biases [24]. Specifically, many of these methods perform poorly when there are a high proportion of individuals with data below the LOD for a particular analyte. In addition, methods which rely on parametric models (i.e. multiple imputations) result in overly narrow confidence intervals when the model is miss-specified [33]. We used a new method based on cox regression to identify associations between immune biomarkers and preeclampsia [33]. This method developed by Dinse et al., was shown to be a valid approach to handle measurements below the LOD especially when analytes with a high proportion of non-detects were present. First, measurements which are left censored (lower detection limits) are reversed to right censored data. Cox regression was then utilized treating the reversed scale for each analyte as an outcome and preeclampsia as the independent variable. The biomarker is treated as censored data and the hazard ratio for a binary health outcome is interpreted as an adjusted odds ratio. We examined associations between baseline demographics (maternal age, socioeconomic status, body mass index, race, and smoking status), pregnancy outcome data (preterm birth, small for gestational age) and preeclampsia using logistic regression. Maternal age, marital status, socioeconomic status (SES), race, smoking, BMI and gestational age at blood draw were considered as potential covariates. Maternal age, SES, BMI, fetal sex, gestational age at blood draw and smoking were included in the final model as they changed the effect size by more than 10%. All analyses were repeated for term and preterm preeclampsia. To account for multiple comparisons, we used a resampling-style step-down method that incorporates correlation and distributional characteristics of the dataset (as is done with permutations) while maintaining control of the family-wise error rate [34]. SAS V9.2 (Cary, NC) was used for analyses.
Results
Table 1 shows that compared to normotensive women, preeclamptic women had a lower maternal age (β=−0.07, p=0.0009), were more likely to be single [odds ratio (OR) 1.8, 95% confidence interval (CI) 1.2–2.5], more likely to have a mid (OR 2.3, 95% CI 1.3–3.8) or low (OR 3.7, 95% CI 2.2–6.4) SES, less likely to smoke during pregnancy (OR 0.6, 95% CI 0.4–0.9), have a higher BMI (β=0.09, p=0.0035) and more likely to have a small for gestational age baby (OR 1.9, 95% CI 1.2–3.2).
Table 1.
Variable | Controls N=432 |
All Cases N=216 |
a Effect estimates, 95% confidence intervals and p-values |
---|---|---|---|
| |||
%(n) | %(n) | ||
| |||
Maternal Age | |||
Median (IQR) | 20.0 (5.0) | 19.0 (5.0) | β = −0.07, p=0.0009 |
| |||
Marital Status | |||
Married | 335 (77.5) | 143 (66.2) | Reference |
Single | 97 (22.5) | 73 (33.8) | 1.8 (1.2 – 2.5) |
| |||
Socio - economic status | |||
Low | 115 (26.9) | 91 (42.5) | 3.7 (2.2 – 6.4) |
Mid | 213 (49.9) | 102 (47.7) | 2.3 (1.3 – 3.8) |
High | 99 (23.2) | 21 (9.8) | Reference |
| |||
Maternal race | |||
White | 274 (63.6) | 125 (57.9) | Reference |
Non-white | 157 (36.4) | 91 (42.1) | 1.3 (0.9 – 1.8) |
| |||
Smoking status | |||
Non-smoker | 247 (57.2) | 147 (68.1) | Reference |
Smoker | 185 (42.8) | 69 (31.9) | 0.6 (0.4 – 0.9) |
| |||
Body mass index | |||
Median (IQR) | 20.8 (3.2) | 21.3 (4.0) | β = 0.09, p=0.0035 |
| |||
PTB <37 wks | |||
No | 376 (87.7) | 178 (84.4) | Reference |
Yes | 53 (12.3) | 33 (15.6) | 1.3 (0.8 – 2.1) |
| |||
PTB <34 wks | |||
No | 408 (95.1) | 198 (93.8) | Reference |
Yes | 21 (4.9) | 13 (6.2) | 1.3 (0.6 – 2.6) |
| |||
Fetal Sex | |||
Male | 221 (51.2) | 102 (47.2) | 1.2 (0.8 – 1.6) |
Female | 211 (48.8) | 114 (52.8) | Reference |
| |||
SGA 5th percentile | |||
No | 382 (91.4) | 171 (84.7) | Reference |
Yes | 36 (8.6) | 31 (15.3) | 1.9 (1.2 – 3.2) |
| |||
SGA 10th percentile | |||
No | 339 (81.1) | 153 (75.7) | Reference |
Yes | 79 (18.9) | 49 (24.3) | 1.4 (0.9 – 2.1) |
Determined by logistic regression. Continuous variables display β, and binary or categorical variables odds ratios.
Among women sampled in the first trimester, preeclamptic women have higher median levels of RANTES, IL4r, IL6, IL12, IL1β, IL10 and TGFβ compared to normotensive women (Table 2). Among women sampled in the second trimester, preeclamptic women had lower median values of IL4r, MIP, IL1β, IL10, and TGFβ compared to normotensive women. In contrast, women with preeclampsia had higher median levels of MIF, IL6, and IL8.
Table 2.
First Trimester N=242 | Second Trimester N=392 | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Normotensive N=178 |
Preeclampsia N=64 |
Normotensive N=233 |
Preeclampsia N=144 |
|||||||||
Biomarker | Mean Pg/ml |
Median Pg/ml |
Range Pg/ml |
Mean Pg/ml |
Median Pg/ml |
Range Pg/ml |
Mean Pg/ml |
Median Pg/ml |
Range Pg/ml |
Mean Pg/ml |
Median Pg/ml |
Range Pg/ml |
RANTES | 47.4 | 44.0 | 18.3–160.0 | 52.4 | 45.9 | 26.0–160.0 | 46.6 | 43.0 | 2.0–160.0 | 49.4 | 43.4 | 23.6–160.0 |
IL4r | 338.9 | 235.0 | 4.0–4000.0 | 507.2 | 247.0 | 4.0–4000.0 | 385.6 | 266.0 | 4.0–4000.0 | 338.3 | 211.0 | 4.0–4000.0 |
MIF | 23.5 | 17.0 | 0.1–100.0 | 21.8 | 16.3 | 0.1–100.0 | 25.8 | 15.4 | 0.05–101.0 | 27.6 | 19.2 | 0.1–100.0 |
MIP | 336.4 | 193.0 | 10.0–7792 | 234.8 | 195.5 | 10.0–1343 | 306.1 | 219.0 | 10.0–4223 | 288.9 | 172.0 | 10.0–3300 |
TNFR1 | 0.78 | 0.51 | 0.08–27.4 | 0.6 | 0.5 | 0.1–2.5 | 0.68 | 0.53 | 0.08–9.3 | 0.86 | 0.5 | 0.1–25.2 |
TNFα | 48.5 | 4.0 | 4.0–754.0 | 40.2 | 4.0 | 4.0–467.0 | 53.4 | 4.0 | 4.0–930.0 | 50.8 | 4.0 | 4.0–599.0 |
IL6 | 190.1 | 63.0 | 4.0–4000.0 | 413.5 | 88.5 | 4.0–4000.0 | 213.1 | 71.0 | 4.0–4000.0 | 182.8 | 73.0 | 4.0–4000.0 |
IL4 | 67.1 | 21.0 | 4.0–2063.0 | 72.7 | 25.0 | 4.0–1064.0 | 77.6 | 21.0 | 4.0–2538.0 | 70.8 | 19.0 | 4.0–1694.0 |
IL5 | 15.3 | 12.0 | 4.0–78.0 | 15.6 | 13.5 | 4.0–65.0 | 17.8 | 11.0 | 4.0–244.0 | 15.2 | 11.0 | 4.0–163.0 |
IL12 | 18.6 | 4.0 | 4.0–1218.0 | 55.9 | 8.5 | 4.0–1508.0 | 16.2 | 4.0 | 4.0–276.0 | 11.9 | 4.0 | 4.0–298.0 |
IL1α | 4.4 | 4.0 | 4.0–29.0 | 4.0 | 4.0 | 4.0–4.0 | 7.3 | 4.0 | 4.0–705.0 | 9.5 | 4.0 | 4.0–745.0 |
IL1β | 21.0 | 9.0 | 4.0–1489.0 | 15.1 | 10.0 | 4.0–98.0 | 19.8 | 10.0 | 4.0–373.0 | 23.3 | 9.0 | 4.0–1634.0 |
IL8 | 201.7 | 21.5 | 4.0–4000.0 | 180.8 | 13.0 | 4.0–1458.0 | 188.0 | 15.0 | 4.0–4000.0 | 190.6 | 20.5 | 4.0–4000.0 |
IL10 | 83.7 | 40.0 | 4.0–801.0 | 92.0 | 44.5 | 4.0–743.0 | 87.4 | 46.0 | 4.0–891.0 | 79.4 | 41.5 | 3.0–836.0 |
IFNγ | 36.3 | 4.0 | 4.0–4000.0 | 33.7 | 4.0 | 4.0–902.0 | 27.3 | 4.0 | 4.0–2117.0 | 36.8 | 4.0 | 4.0–2450.0 |
TGFβ | 804.0 | 604.0 | 39.0–5501 | 897.8 | 718.0 | 39.0–3468 | 952.4 | 872.0 | 39.0–3765 | 814.9 | 682.5 | 39.0–4882 |
TFNβ | 66.4 | 10.0 | 10.0–2364 | 65.2 | 10.0 | 10.0–1658 | 94.7 | 10.0 | 10.0–6821 | 114.5 | 10.0 | 10.0–10000 |
In first trimester samples, only IL-12 was significantly associated with preeclampsia (HR 1.6, 95% CI 1.1–2.6; p=0.0255) after adjustments (Table 3). IFNγ (HR 3.5, 95% CI 1.4–8.5; p=0.0063), IL1β (HR 3.6, 95% CI 1.7–7.6; p=0.0006), IL5 (HR 2.1, 95% CI 1.0–4.2; p=0.0422), and TNFr (HR 2.0, 95% CI 1.0–3.6; p=0.0460) were significantly associated with preterm preeclampsia. The association between IL1β and preterm preeclampsia remained significant after correcting for multiple comparisons.
Table 3.
All PE (n=64) a HR (95% CI) |
Raw p-value |
b Adjusted p-value |
Preterm PE (n=10) a HR (95% CI) |
Raw p-value |
b Adjusted p-value |
Term PE (n=54) a HR (95% CI) |
Raw p-value |
b Adjusted p-value |
|
---|---|---|---|---|---|---|---|---|---|
IFNγ | 1.2 (0.7–2.2) | 0.4853 | 0.9493 | 3.5 (1.4–8.5) | 0.0063 | 0.1134 | 0.9 (0.5–1.8) | 0.8645 | 0.9493 |
IL1β | 1.2 (0.8–1.8) | 0.4928 | 0.9493 | 3.6 (1.7–7.6) | 0.0006 | 0.0216 | 0.9 (0.6–1.5) | 0.8240 | 0.9493 |
IL4 | 1.0 (0.7–1.5) | 0.8036 | 0.9493 | 1.2 (0.6–2.6) | 0.4880 | 0.9493 | 1.0 (0.7–1.5) | 0.9493 | 0.9493 |
IL4r | 1.1 (0.8–1.6) | 0.4645 | 0.9493 | 1.5 (0.7–2.8) | 0.2754 | 0.8019 | 1.1 (0.8–1.5) | 0.6458 | 0.9493 |
IL5 | 1.0 (0.7–1.6) | 0.6725 | 0.9493 | 2.1 (1.0–4.2) | 0.0422 | 0.5064 | 0.9 (0.6–1.5) | 0.8217 | 0.9493 |
IL8 | 1.0 (0.8–1.4) | 0.8410 | 0.9493 | 0.8 (0.4–1.8) | 0.6351 | 0.5885 | 1.0 (0.8–1.4) | 0.8977 | 0.9493 |
IL6 | 1.3 (0.9–1.9) | 0.1431 | 0.6244 | 1.5 (0.8–3.0) | 0.1750 | 0.6886 | 1.3 (0.9–1.8) | 0.2543 | 0.7629 |
IL6r | 0.9 (0.7–1.4) | 0.9445 | 0.9493 | 1.0 (0.5–2.0) | 0.8894 | 0.9493 | 1.0 (0.5–2.0) | 0.8981 | 0.9493 |
IL10 | 1.1 (0.8–1.7) | 0.4894 | 0.9493 | 1.6 (0.8–3.2) | 0.2060 | 0.7063 | 1.1 (0.7–1.6) | 0.7499 | 0.9493 |
IL12 | 1.6 (1.1–2.6) | 0.0255 | 0.5064 | 2.2 (1.0–5.0) | 0.0589 | 0.5774 | 1.6 (1.0–2.5) | 0.0655 | 0.5648 |
MIF | 1.0 (0.7–1.4) | 0.8925 | 0.9493 | 0.8 (0.4–1.7) | 0.5518 | 0.9493 | 1.0 (0.7–1.5) | 0.9201 | 0.9493 |
MIP | 1.2 (0.8–1.7) | 0.3866 | 0.9278 | 1.7 (0.9–3.4) | 0.0973 | 0.5838 | 1.1 (0.7–1.6) | 0.6814 | 0.9493 |
TGFβ | 1.2 (0.9–1.7) | 0.2741 | 0.7739 | 1.5 (0.8–3.1) | 0.2361 | 0.7555 | 1.2 (0.8–1.7) | 0.4155 | 0.9493 |
TNFα | 0.9 (0.6–1.5) | 0.7802 | 0.9493 | 0.9 (0.4–2.5) | 0.8184 | 0.9493 | 0.9 (0.6–1.6) | 0.8230 | 0.9493 |
TNFβ | 1.1 (0.6–2.1) | 0.7305 | 0.9493 | 2.1 (0.7–5.9) | 0.1786 | 0.6886 | 1.0 (0.5–1.9) | 0.9341 | 0.9493 |
TNFr | 0.9 (0.6–1.2) | 0.3470 | 0.8766 | 2.0 (1.0–3.6) | 0.0460 | 0.5462 | 0.8 (0.5–1.1) | 0.1138 | 0.5885 |
Hazard ratio (HR) and 95% confidence intervals (CI) were calculated using cox regression. Hazard ratios can be interpreted as adjusted odds ratios (Dinse et al.) All models were adjusted for maternal age, fetal sex, BMI, SES, gestational age at blood draw and smoking
Adjusted for multiple comparisons
We observed different trends in second trimester samples (Table 4). Decreased IL1β (HR 0.7, 95% CI 0.5–0.9; p=0.0180) and TGFβ (HR 0.8, 95% CI 0.6–0.9; p=0.0292) were significantly associated with all preeclampsia cases. No associations were found with preterm preeclampsia. No other significant associations were observed. Results were not significant after correction for multiple comparisons.
Table 4.
All PE (n=144) a HR (95% CI) |
Raw p-value |
b Adjusted p-value |
Preterm PE (n=27) a HR (95% CI) |
Raw p-value |
b Adjusted p-value |
Term PE (n=117) a HR (95% CI) |
Raw p-value |
b Adjusted p-value |
|
---|---|---|---|---|---|---|---|---|---|
IFNγ | 0.9 (0.6–1.4) | 0.5997 | 0.9687 | 0.8 (0.3–1.9) | 0.5965 | 0.9687 | 0.9 (0.6–1.4) | 0.7395 | 0.9687 |
IL1β | 0.7 (0.5–0.9) | 0.0180 | 0.5350 | 0.6 (0.4–1.1) | 0.0860 | 0.7559 | 0.7 (0.5–0.9) | 0.0454 | 0.5350 |
IL4 | 0.8 (0.6–1.1) | 0.2394 | 0.6760 | 1.0 (0.6–1.6) | 0.9968 | 0.9968 | 0.8 (0.6–1.1) | 0.1743 | 0.6436 |
IL4r | 1.0 (0.6–1.1) | 0.1089 | 0.5350 | 0.9 (0.6–1.4) | 0.6518 | 0.9687 | 0.8 (0.6–1.0) | 0.0962 | 0.5350 |
IL5 | 1.1 (0.8–1.4) | 0.5383 | 0.9687 | 1.5 (0.9–2.3) | 0.1226 | 0.5350 | 1.0 (0.8–1.4) | 0.8977 | 0.9968 |
IL8 | 1.0 (0.7–1.3) | 0.9674 | 0.9968 | 0.6 (0.4–1.2) | 0.2218 | 0.6760 | 1.1 (0.8–1.4) | 0.6390 | 0.9687 |
IL6 | 1.1 (0.8–1.3) | 0.9510 | 0.9968 | 1.1 (0.7–1.7) | 0.7011 | 0.9687 | 1.0 (0.7–1.3) | 0.9292 | 0.9968 |
IL6r | 1.0 (0.8–1.3) | 0.8437 | 0.9968 | 1.1 (0.7–1.8) | 0.5325 | 0.9687 | 1.0 (0.8–1.3) | 0.9915 | 0.9968 |
IL10 | 0.8 (0.6–1.1) | 0.1582 | 0.6328 | 1.0 (0.6–1.6) | 0.9877 | 0.9968 | 0.8 (0.6–1.1) | 0.1074 | 0.5350 |
IL12 | 0.8 (0.6–1.2) | 0.2260 | 0.6760 | 1.1 (0.6–1.9) | 0.8394 | 0.9968 | 0.9 (0.6–1.2) | 0.4044 | 0.9243 |
MIF | 1.1 (0.8–1.3) | 0.6853 | 0.9687 | 0.9 (0.6–1.4) | 0.5684 | 0.9687 | 1.1 (0.8–1.4) | 0.5944 | 0.9687 |
MIP | 0.9 (0.7–1.1) | 0.3212 | 0.7709 | 0.8 (0.5–1.3) | 0.2384 | 0.6760 | 0.9 (0.7–1.2) | 0.4647 | 0.9687 |
TGFβ | 0.8 (0.6–0.9) | 0.0292 | 0.5350 | 0.7 (0.4–1.1) | 0.1173 | 0.5350 | 0.8 (0.6–1.0) | 0.0642 | 0.5350 |
TNFα | 1.1 (0.8–1.5) | 0.7669 | 0.9687 | 0.9 (0.5–1.7) | 0.7224 | 0.9687 | 1.1 (0.8–1.5) | 0.6249 | 0.9687 |
TNFβ | 0.7 (0.4–1.1) | 0.1158 | 0.5350 | 0.4 (0.1–1.2) | 0.1028 | 0.5350 | 0.8 (0.5–1.2) | 0.2711 | 0.7229 |
TNFr | 1.0 (0.8–1.3) | 0.7468 | 0.9687 | 1.0 (0.6–1.6) | 0.9493 | 0.9968 | 1.0 (0.8–1.3) | 0.7300 | 0.9687 |
Hazard ratio (HR) and 95% confidence intervals (CI) were calculated using cox regression. Hazard ratios can be interpreted as adjusted odds ratios (Dinse et al.) All models were adjusted for maternal age, BMI, SES, gestational age at blood draw and smoking
Adjusted for multiple comparisons
As our second trimester results differed from the first trimester, we compared maternal characteristics between women with first trimester samples and women with second trimester samples. Compared to first trimester controls, second trimester controls were younger, more likely to smoke, had a lower SES, and were more likely to be African American (Chi-square test p<0.05 for all). Second trimester cases were similar to first trimester cases except that they were more likely to be African American and less likely to have a SGA baby (10th percentile based on maternal characteristics) (Chi-square test p<0.05).
As an exploratory analysis we stratified analyses by race to determine if potential effect modification exist. We also explored differences between preeclampsia with and without SGA. In the second trimester, TNFβ (HR 0.6, 95% CI 0.3–0.9) was associated with preeclampsia without SGA and there was no difference by race. Similarly, IL1β was associated with preeclampsia without SGA (HR 0.6, 95% CI 0.5–0.8) and effect estimates were similar among races. There was some effect modification by race present for IL6 which was associated with preeclampsia in white (HR 1.7, 95% CI 1.1–2.9) but not African American women (HR 0.7, 95% CI 0.5–1.3).
Discussion
We found that IL1β was associated with preterm preeclampsia in the first trimester. Our results are similar to a study of 70 women reporting that first trimester IL1β is a significant predictor of early onset preeclampsia [14]. IL1β has been suggested to be necessary for immune-tolerance at the maternal-fetal interface through modulation of the nuclear factor kappa-B pathway [35]. Alteration in immune-tolerance could adversely effect placentation. Indeed, a recent placental gene expression study identified an immunological preeclampsia subtype that correlates with placental dysfunction and poor fetal growth [36]. Our study measured systemic markers which may not represent placental immune dysfunction. Still, severe subtypes of preeclampsia (i.e. early-onset, poor fetal growth) are hypothesized to be associated with abnormal placentation while “maternal” preeclampsia may occur despite normal placental function [36, 37]. Placental dysfunction increases syncytiotrophoblast microvesicle production leading to exaggerated systemic inflammation [6]. In fact, treatment of peripheral blood mononuclear cells by first trimester microvesicles has been shown to increase IL1β secretion [38]. Thus, increased systemic IL1β in the first trimester may indicate an immunological preeclampsia subtype perhaps mediated by abnormal placentation and increased placental microvesicles.
Our results were conflicting in second trimester samples where IL1β was lower in preeclamptic women. However, these findings replicate our previous study where mid-pregnancy IL1β was associated with decreased odds of preeclampsia in the Danish National Birth Cohort [15]. The results are also consistent with another study of serum IL1β at 34 weeks gestation [18]. Still, the reasons for these differing results in first and second trimester samples are not entirely clear. Placental microvesicles are hypothesized to drive systemic inflammation. In healthy pregnancy, microvesicle production of IL1β from peripheral blood mononuclear cells is not observed at term [38]. In fact, IL1β is downregulated by macrophages following phagocytosis of placental microvesicles at term [39]. This may suggest that cytokine production via placental microvesicles is increased in early pregnancy but not in later gestations. Conversely, microvesicles from preeclamptic pregnancies are reported to increase IL1β production in the first trimester and at term [38]. However, investigations among term preeclamptic women may be biased by MgS04 treatment which increases IL1β [40]. Longitudinal profiles of systemic immune markers during pregnancy are limited. IL1β has been shown to significantly decrease from the first to second trimester before increasing again at delivery [41, 42]. One longitudinal study which began sampling at 18 weeks gestation until term, found no associations between IL1β and preeclampsia [43]. Thus, systemic IL1β levels and associations with preeclampsia are not consistent across pregnancy. Our results may suggest that increased first trimester systemic IL1β is downregulated later in pregnancy possible due to a compensatory mechanism.
We observed differences in maternal characteristics between women sampled in the first and second trimesters. Controls sampled in the second trimester were more likely to be young African Americans, smoke and have a low SES compared to first trimester controls. Second trimester cases were also more likely to be African American but less likely to have a SGA baby compared to first trimester cases. Maternal characteristics may impact systemic immune markers [44]. Thus, second trimester controls may have been more likely to have elevated immune markers via unmeasured maternal factors biasing results towards the null. Lastly, it is also possible that a decrease in IL1β in the second trimester is capturing a different preeclampsia subtype [36]. Our exploratory analyses did show that in the second trimester IL1β was significantly decreased in preeclampsia without SGA but displayed trends towards increased levels in preeclampsia with SGA.
We found no other significant associations after correction for multiple comparisons. In a previous investigation, we found that mid-trimester IL-6 and TNFβ were associated with term preeclampsia [15]. However, the current investigation could not replicate these findings. The conflicting results may be due to differences in the study populations (contemporary Danish Cohort vs. older American cohort). In our exploratory analysis, IL6 was associated with preeclampsia but only in white women. Overall, we found few associations between systemic immune markers and preeclampsia in second trimester samples which is consistent with other mid-trimester investigations [11, 12]. Thus, second trimester serum immune markers may have little use for prediction of preeclampsia.
Our study included a large sample size and blood sample collection prior to the third trimester of pregnancy. We were able to investigate both first and second trimester samples but were limited by a single time point for each subject. We did not have data on time of diagnosis of preeclampsia. However, samples were collected prior to 20 weeks of gestation (range 5–18 weeks). Still, we cannot rule out subclinical disease prior to sampling. We acknowledge that the age of the CPP cohort is a limitation and may affect the stability of our immune markers. Analytes measured using CPP data, including cytokines, have shown to be stable in previous studies. Compared to our study within the DNBC [15], most immune markers measured in the CPP cohort have similar proportions of non-detectable levels. IL1β levels do appear to be lower in the CPP compared to the DNBC. However, we observed the same association with second trimester IL1β in both studies and these cohorts are not comparable in many ways due to differences in demographics.
We found that IL-1β was significantly associated with preterm preeclampsia in the first trimester. Elevated IL-1β in early pregnancy may indicate a subtype of preeclampsia. However, these associations were not observed in the second trimester. Longitudinal changes in IL1β in relation to preeclampsia subtypes may be warranted. Overall, no single immune biomarker is likely a strong predictor for preeclampsia or preeclampsia subtypes, particularly in the second trimester. Investigations which combine several immune markers, biomarkers from pathways which may induce inflammation and clinical data may be useful to define an immunological subtype of preeclampsia.
Highlights.
Serum IL1β measured in the first trimester was associated with preterm preeclampsia after correction for multiple comparisons.
IFNγ, IL5, and TNFr measured in the first trimester were also associated with preterm preeclampsia but not after correction for multiple comparisons.
Analysis among second trimester samples revealed no significant associations after adjustments.
Acknowledgments
None.
Funding: R01HD048669 from the National Institute of Allergy and Infectious Diseases (CH).
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Disclosure: No conflicts of interest to disclose
References
- 1.Hutcheon JA, Lisonkova S, Joseph KS. Epidemiology of pre-eclampsia and the other hypertensive disorders of pregnancy. Best Pract Res Clin Obstet Gynaecol. 2011;25(4):391–403. doi: 10.1016/j.bpobgyn.2011.01.006. [DOI] [PubMed] [Google Scholar]
- 2.O. American College of, Gynecologists, P. Task Force on Hypertension in, Hypertension in pregnancy. Report of the American College of Obstetricians and Gynecologists’ Task Force on Hypertension in Pregnancy. Obstetrics and gynecology. 2013;122(5):1122–31. doi: 10.1097/01.AOG.0000437382.03963.88. [DOI] [PubMed] [Google Scholar]
- 3.Mincheva-Nilsson L, Baranov V. Placenta-derived exosomes and syncytiotrophoblast microparticles and their role in human reproduction: immune modulation for pregnancy success. Am J Reprod Immunol. 2014;72(5):440–57. doi: 10.1111/aji.12311. [DOI] [PubMed] [Google Scholar]
- 4.Tannetta DS, Dragovic RA, Gardiner C, Redman CW, Sargent IL. Characterisation of syncytiotrophoblast vesicles in normal pregnancy and pre-eclampsia: expression of Flt-1 and endoglin. PLoS One. 2013;8(2):e56754. doi: 10.1371/journal.pone.0056754. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Roberts JM, Lain KY. Recent Insights into the pathogenesis of pre-eclampsia. Placenta. 2002;23(5):359–72. doi: 10.1053/plac.2002.0819. [DOI] [PubMed] [Google Scholar]
- 6.Redman CW, Sargent IL. Latest advances in understanding preeclampsia. Science. 2005;308(5728):1592–4. doi: 10.1126/science.1111726. [DOI] [PubMed] [Google Scholar]
- 7.Szarka A, Rigo J, Jr, Lazar L, Beko G, Molvarec A. Circulating cytokines, chemokines and adhesion molecules in normal pregnancy and preeclampsia determined by multiplex suspension array. BMC immunology. 2010;11:59. doi: 10.1186/1471-2172-11-59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Madazli R, Aydin S, Uludag S, Vildan O, Tolun N. Maternal plasma levels of cytokines in normal and preeclamptic pregnancies and their relationship with diastolic blood pressure and fibronectin levels. Acta Obstet Gynecol Scand. 2003;82(9):797–802. doi: 10.1034/j.1600-0412.2003.00206.x. [DOI] [PubMed] [Google Scholar]
- 9.Jonsson Y, Ruber M, Matthiesen L, Berg G, Nieminen K, Sharma S, Ernerudh J, Ekerfelt C. Cytokine mapping of sera from women with preeclampsia and normal pregnancies. Journal of reproductive immunology. 2006;70(1–2):83–91. doi: 10.1016/j.jri.2005.10.007. [DOI] [PubMed] [Google Scholar]
- 10.Kumar A, Begum N, Prasad S, Agarwal S, Sharma S. IL-10, TNF-alpha & IFN-gamma: potential early biomarkers for preeclampsia. Cell Immunol. 2013;283(1–2):70–4. doi: 10.1016/j.cellimm.2013.06.012. [DOI] [PubMed] [Google Scholar]
- 11.Djurovic S, Clausen T, Wergeland R, Brosstad F, Berg K, Henriksen T. Absence of enhanced systemic inflammatory response at 18 weeks of gestation in women with subsequent pre-eclampsia. BJOG. 2002;109(7):759–64. doi: 10.1111/j.1471-0528.2002.01330.x. [DOI] [PubMed] [Google Scholar]
- 12.Mundim GJ, Paschoini MC, Araujo E, Junior, Da Silva Costa F, Rodrigues V., Junior Assessment of angiogenesis modulators in pregnant women with pre-eclampsia: a case-control study. Arch Gynecol Obstet. 2016;293(2):369–75. doi: 10.1007/s00404-015-3823-x. [DOI] [PubMed] [Google Scholar]
- 13.Lockwood CJ, Huang SJ, Chen CP, Huang Y, Xu J, Faramarzi S, Kayisli O, Kayisli U, Koopman L, Smedts D, Buchwalder LF, Schatz F. Decidual cell regulation of natural killer cell-recruiting chemokines: implications for the pathogenesis and prediction of preeclampsia. Am J Pathol. 2013;183(3):841–56. doi: 10.1016/j.ajpath.2013.05.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Siljee JE, Wortelboer EJ, Koster MP, Imholz S, Rodenburg W, Visser GH, de Vries A, Schielen PC, Pennings JL. Identification of interleukin-1 beta, but no other inflammatory proteins, as an early onset pre-eclampsia biomarker in first trimester serum by bead-based multiplexed immunoassays. Prenat Diagn. 2013;33(12):1183–8. doi: 10.1002/pd.4219. [DOI] [PubMed] [Google Scholar]
- 15.Taylor BD, Tang G, Ness RB, Olsen J, Hougaard DM, Skogstrand K, Roberts JM, Haggerty CL. Mid-pregnancy circulating immune biomarkers in women with preeclampsia and normotensive controls. Pregnancy Hypertens. 2016;6(1):72–8. doi: 10.1016/j.preghy.2015.11.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Perez-Sepulveda A, Torres MJ, Khoury M, Illanes SE. Innate immune system and preeclampsia. Front Immunol. 2014;5:244. doi: 10.3389/fimmu.2014.00244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Redman CW, Sargent IL. Pre-eclampsia, the placenta and the maternal systemic inflammatory response–a review. Placenta. 2003;24(Suppl A):S21–7. doi: 10.1053/plac.2002.0930. [DOI] [PubMed] [Google Scholar]
- 18.Mosimann B, Wagner M, Poon LC, Bansal AS, Nicolaides KH. Maternal serum cytokines at 30–33 weeks in the prediction of preeclampsia. Prenatal diagnosis. 2013;33(9):823–30. doi: 10.1002/pd.4129. [DOI] [PubMed] [Google Scholar]
- 19.Haggerty CL, Klebanoff MA, Panum I, Uldum SA, Bass DC, Olsen J, Roberts JM, Ness RB. Prenatal infection increases the risk of preeclampsia. Pregnancy hypertension. 2013;3(3):151–154. doi: 10.1016/j.preghy.2013.03.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Niswander K. The Collaborative Perinatal Study of the National Institute of Neurological Diseases and Stroke: The Women and Their Pregnancies. Philadelphia: PA WB Saybders; 1972. [Google Scholar]
- 21.Hardy JB. The Collaborative Perinatal Project: lessons and legacy. Ann Epidemiol. 2003;13(5):303–11. doi: 10.1016/s1047-2797(02)00479-9. [DOI] [PubMed] [Google Scholar]
- 22.Myrianthopoulos NC, French KS. An application of the U.S. Bureau of the Census socioeconomic index to a large, diversified patient population. Soc Sci Med. 1968;2(3):283–99. doi: 10.1016/0037-7856(68)90004-8. [DOI] [PubMed] [Google Scholar]
- 23.Skogstrand K, Thorsen P, Norgaard-Pedersen B, Schendel DE, Sorensen LC, Hougaard DM. Simultaneous measurement of 25 inflammatory markers and neurotrophins in neonatal dried blood spots by immunoassay with xMAP technology. Clinical chemistry. 2005;51(10):1854–66. doi: 10.1373/clinchem.2005.052241. [DOI] [PubMed] [Google Scholar]
- 24.Taylor BD, Holzman CB, Fichorova RN, Tian Y, Jones NM, Fu W, Senagore PK. Inflammation biomarkers in vaginal fluid and preterm delivery. Human reproduction. 2013;28(4):942–52. doi: 10.1093/humrep/det019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Aye IL, Lager S, Ramirez VI, Gaccioli F, Dudley DJ, Jansson T, Powell TL. Increasing maternal body mass index is associated with systemic inflammation in the mother and the activation of distinct placental inflammatory pathways. Biology of reproduction. 2014;90(6):129. doi: 10.1095/biolreprod.113.116186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Martins TB, Pasi BM, Pickering JW, Jaskowski TD, Litwin CM, Hill HR. Determination of cytokine responses using a multiplexed fluorescent microsphere immunoassay. American journal of clinical pathology. 2002;118(3):346–53. doi: 10.1309/N0T6-C56B-GXB2-NVFB. [DOI] [PubMed] [Google Scholar]
- 27.Whitcomb BW, Schisterman EF, Luo X, Chegini N. Maternal serum granulocyte colony-stimulating factor levels and spontaneous preterm birth. J Womens Health (Larchmt) 2009;18(1):73–8. doi: 10.1089/jwh.2008.0883. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Whitcomb BW, Schisterman EF, Klebanoff MA, Baumgarten M, Rhoton-Vlasak A, Luo X, Chegini N. Circulating chemokine levels and miscarriage. Am J Epidemiol. 2007;166(3):323–31. doi: 10.1093/aje/kwm084. [DOI] [PubMed] [Google Scholar]
- 29.Stroud LR, Solomon C, Shenassa E, Papandonatos G, Niaura R, Lipsitt LP, Lewinn K, Buka SL. Long-term stability of maternal prenatal steroid hormones from the National Collaborative Perinatal Project: still valid after all these years. Psychoneuroendocrinology. 2007;32(2):140–50. doi: 10.1016/j.psyneuen.2006.11.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Whitcomb BW, Schisterman EF, Klebanoff MA, Baumgarten M, Luo X, Chegini N. Circulating levels of cytokines during pregnancy: thrombopoietin is elevated in miscarriage. Fertil Steril. 2008;89(6):1795–802. doi: 10.1016/j.fertnstert.2007.05.046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Myatt L, Redman CW, Staff AC, Hansson S, Wilson ML, Laivuori H, Poston L, Roberts JM, C Global Pregnancy Strategy for standardization of preeclampsia research study design. Hypertension. 2014;63(6):1293–301. doi: 10.1161/HYPERTENSIONAHA.113.02664. [DOI] [PubMed] [Google Scholar]
- 32.Uh HW, Hartgers FC, Yazdanbakhsh M, Houwing-Duistermaat JJ. Evaluation of regression methods when immunological measurements are constrained by detection limits. BMC immunology. 2008;9:59. doi: 10.1186/1471-2172-9-59. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Dinse GE, Jusko TA, Ho LA, Annam K, Graubard BI, Hertz-Picciotto I, Miller FW, Gillespie BW, Weinberg CR. Accommodating measurements below a limit of detection: a novel application of Cox regression. Am J Epidemiol. 2014;179(8):1018–24. doi: 10.1093/aje/kwu017. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Westfall PH, Young SS. Resampling-based multiple testing: Examples and Methods for p-value adjustment. John Wiley and Sons; New York: 1993. [Google Scholar]
- 35.Geisert R, Fazleabas A, Lucy M, Mathew D. Interaction of the conceptus and endometrium to establish pregnancy in mammals: role of interleukin 1beta. Cell and tissue research. 2012;349(3):825–38. doi: 10.1007/s00441-012-1356-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Leavey K, Benton SJ, Grynspan D, Kingdom JC, Bainbridge SA, Cox BJ. Unsupervised Placental Gene Expression Profiling Identifies Clinically Relevant Subclasses of Human Preeclampsia. Hypertension. 2016 doi: 10.1161/HYPERTENSIONAHA.116.07293. [DOI] [PubMed] [Google Scholar]
- 37.Staff AC, Benton SJ, von Dadelszen P, Roberts JM, Taylor RN, Powers RW, Charnock-Jones DS, Redman CW. Redefining preeclampsia using placenta-derived biomarkers. Hypertension. 2013;61(5):932–42. doi: 10.1161/HYPERTENSIONAHA.111.00250. [DOI] [PubMed] [Google Scholar]
- 38.Holder BS, Tower CL, Jones CJ, Aplin JD, Abrahams VM. Heightened pro-inflammatory effect of preeclamptic placental microvesicles on peripheral blood immune cells in humans. Biol Reprod. 2012;86(4):103. doi: 10.1095/biolreprod.111.097014. [DOI] [PubMed] [Google Scholar]
- 39.Abumaree MH, Chamley LW, Badri M, El-Muzaini MF. Trophoblast debris modulates the expression of immune proteins in macrophages: a key to maternal tolerance of the fetal allograft? J Reprod Immunol. 2012;94(2):131–41. doi: 10.1016/j.jri.2012.03.488. [DOI] [PubMed] [Google Scholar]
- 40.Amash A, Holcberg G, Sapir O, Huleihel M. Placental secretion of interleukin-1 and interleukin-1 receptor antagonist in preeclampsia: effect of magnesium sulfate. Journal of interferon & cytokine research: the official journal of the International Society for Interferon and Cytokine Research. 2012;32(9):432–41. doi: 10.1089/jir.2012.0013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Christian LM, Porter K. Longitudinal changes in serum proinflammatory markers across pregnancy and postpartum: effects of maternal body mass index. Cytokine. 2014;70(2):134–40. doi: 10.1016/j.cyto.2014.06.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Ferguson KK, McElrath TF, Chen YH, Mukherjee B, Meeker JD. Longitudinal profiling of inflammatory cytokines and C-reactive protein during uncomplicated and preterm pregnancy. Am J Reprod Immunol. 2014;72(3):326–36. doi: 10.1111/aji.12265. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Kronborg CS, Gjedsted J, Vittinghus E, Hansen TK, Allen J, Knudsen UB. Longitudinal measurement of cytokines in pre-eclamptic and normotensive pregnancies. Acta Obstet Gynecol Scand. 2011;90(7):791–6. doi: 10.1111/j.1600-0412.2011.01134.x. [DOI] [PubMed] [Google Scholar]
- 44.Curry AE, Vogel I, Skogstrand K, Drews C, Schendel DE, Flanders WD, Hougaard DM, Thorsen P. Maternal plasma cytokines in early- and mid-gestation of normal human pregnancy and their association with maternal factors. J Reprod Immunol. 2008;77(2):152–60. doi: 10.1016/j.jri.2007.06.051. [DOI] [PubMed] [Google Scholar]