Inflammation and Cardiovascular Risk in Women with Preterm Labor

Introduction

Women with a history of preterm delivery have about twice the normal risk of cardiovascular disease (CVD), as described in observational studies, ^{1 –4} although mechanisms underlying this association are not well understood. Components of inflammation are present in women with preterm labor. As infection is the underlying cause of preterm labor and preterm premature rupture of membranes (PPROM) in about 25%–40% of cases, ^5,6 it has been thought that infection is the origin of inflammation in these women. Nevertheless, other findings, such as the presence of leukocyte infiltrate in the cervical and uterine tissues in the absence of infection, ⁷ suggest that preterm labor could be an inflammatory condition even in the absence of infection. This would help to explain why these women are at increased risk of CVD later in life, as inflammation is also an independent predictor of this condition in both men and women. ⁸

Thus, not only infection but also other conditions related to inflammation should be investigated in women with or at risk of preterm labor. Chronic systemic inflammation is significantly associated with some metabolic conditions, such as the metabolic syndrome, that are known risk factors for CVD diseases. ⁹ Hypothetically, the existence of chronic inflammation linked to metabolic risk factors during pregnancy could be the link between preterm labor and a subsequently increased risk of CVD and metabolic diseases. More specifically, dyslipidemia, a known CVD risk factor, has been reported to be related to both spontaneous preterm labor (sPL) ¹⁰ and inflammation, ¹¹ but it is not clear whether or not both associations are connected to each other. Confirmation of these important observations is needed, ideally in prospective studies, along with an exploration of the inflammatory mechanisms common to both sPL and CVD.

The aim of the present study was, first, to test the hypothesis that dyslipidemia and other metabolic CVD risk factors are more prevalent in women with sPL and, second, to evaluate the relationships among these risk factors and both proinflammatory cytokines and markers of endothelial dysfunction.

Material and Methods

This was a prospective study performed during pregnancy on 90 pregnant women allocated into two groups: 40 women with sPL and 50 healthy pregnant women who delivered at term. The study was approved by the local Ethical and Research Committee.

Pregnant women who were admitted to our center with the diagnosis of threatened sPL (between 24 and 36 weeks gestational age) were eligible as cases and were offered participation in the study. Women were excluded for any maternal or fetal condition needing delivery (including symptoms or signs of chorioamnionitis), multiple gestations, PROM, intrauterine fetal demise or suspected lethal fetal anomalies, cervix dilated > 4 cm, treatment with either tocolytics or corticosteroids within 24 hours previously, food intake within 8 hours previously, or failure to consent. Those who accepted signed a written informed consent. Those who subsequently had a positive urine or cervical culture (which were performed in all cases) and those who finally delivered after 36 weeks of gestation were subsequently excluded from the study. Preterm labor was based on the clinical diagnosis of at least four painful uterine contractions per 20 minutes and evidence of cervical change (Bishop score > 6).

The control group comprised 50 healthy pregnant women with noncomplicated pregnancies who were recruited from our antenatal clinic in the third trimester of pregnancy among those with gestational age similar to that of cases. All of them were followed until the end of pregnancy, and none of them experienced either late pregnancy complications or preterm labor, which were considered the exclusion criteria for this group. Initially, 120 consecutive women were recruited, 60 in each group. Subsequently, 20 and 10 women from the preterm labor and the control groups, respectively, were excluded according to the exclusion criteria. Despite these exclusions, the final sample size was higher than that required for the sample size calculation.

Sample size was calculated using data from a pilot study. We calculated that 36 women in each group would be needed to find a significant difference with an α level of 0.05 and a power (1 – β ie; pre-established statistical power of the study) of 80% in the circulating levels of high-density lipoprotein cholesterol (HDL-C) (mean difference 10 mg/dL, with standard deviation [SD] of 15 mg/dL in each group). We chose dyslipidemia as the main variable to calculate sample size, as it had been reported as present in women with sPL. ¹⁰ We initially enrolled 60 women in each group to account for withdrawals and for secondary analysis of data.

Prepregnancy maternal body mass index (BMI) was calculated for each patient (weight in kilograms/height in meters ² ) using the data of height and weight from the clinical records and asking the patients specifically about prepregnancy weight, comparing the answers with the first maternal weight measured in the first trimester of pregnancy and asking again when unexpected values were found. This was the only prepregnancy variable used in this study. We decided against using gestational BMI because not only does this variable represent maternal adiposity but it also is influenced by many other gestational factors, such as fetal growth, placental and amniotic fluid volume, and increased circulating volume.

Blood pressure was taken and recorded at inclusion in the study. Gestational age calculations were based on last menstrual period information and were adjusted using an early ultrasound estimate. In cases of sPL, maternal blood was collected within 2 hours of admission (only if women were in the fasting state as required by the exclusion criteria), allowed to clot and centrifuged for 10 minutes. Glycemia, triglycerides, insulin, total cholesterol, low-density lipoprotein cholesterol (LDL-C), and HDL-C serum levels were measured by standard automated methods. Insulin resistance was evaluated by the homeostasis model assessment (HOMA). For the rest of the biochemical measurements, the sera were frozen at −20°C until assayed. In the control group, maternal blood was collected at the time of admission into the study.

Obesity was defined as BMI≥30 kg/m². In addition, all the women considered as obese who were participating in the study were followed up after delivery, and all of them had an abdominal circumference > 80 cm in the postpartum period. Fasting hyperglycemia in pregnancy was considered when the glycemia was > 105 mg/dL. Two atherogenic indexes (AI): AI1 (total cholesterol/HDL-C) and AI2 (LDL-C/HDL-C) were calculated.

Markers of endothelial dysfunction (soluble vascular cell adhesion molecule [sVCAM], soluble intercellular adhesion molecule [sICAM], myeloperoxidase [MPO], and E-selectin) and cytokine (tumor necrosis factor-alpha [TFN-alpha], interleukin-6 [IL-6], leptin) concentrations were measured using the flow cytometry-based multiplex assay system Luminex MultiAnalyte Profiling Technology (LabMAP, Luminex Corp., Austin, TX). We measured both standards and samples in duplicate. Concentrations were calculated from the standard curves using linear regression. Distribution of variables was checked by both the Kolmogorov-Smirnov test and analysis of the histograms. All the studied variables followed a normal distribution with the exception of IL-6, insulinemia, HOMA, MPO, and ICAM. For these variables, logarithmic transformations were used for statistical analysis. Numerical data are shown as means and SDs. The relationships between variables were analyzed using the Pearson linear correlation coefficient. Forward stepwise (conditional) binary logistic regression was used to analyze the predictive capacity of the studied variables on preterm labor. Differences between women with sPL and those in the control group were studied using Student's t test. The level of significance was set previously at the 95% level (p<0.05).

Results

Maternal age and gestational age were similar in both groups (maternal age 27.65±6.51 vs. 29.64±4.98 years, p=0.10; weeks of gestation 31.27±2.14 vs. 31.56±3.14, gestation, p=0.62 for the sPL and control groups, respectively). The proportion of smokers in both groups was similar. There were only 5 smokers in the sPL group (12.5%) and 4 in the control group (8%) (p=0.50). Number of cigarettes per day was small and also similar in both groups (4.80±3.03 vs. 4.00±3.36, p=0.72). By definition, gestational age at delivery and birth weight were significantly lower in the sPL group than in the control group (33.85±2.60 vs. 39.57±2.25 weeks, p<0.0001; 2398.50±696.87 g vs. 3403.33±425.69 g, p<0.0001).

Anthropometric measurements and metabolic CVD risk factors are shown in Table 1. Maternal weight and BMI were significantly lower in women with sPL (p=0.002 and p=0.01, respectively). Maternal glycemia was higher in women with sPL, but the differences did not reach statistical significance (p=0.10), while insulin levels and insulin sensitivity were similar in both groups. Blood pressure was also similar in both groups. There were no cases of hyperglycemia or hypertension in any of the studied groups. There were no significant differences in triglycerides levels, but all the cholesterol fractions were significantly lower in the sPL group, especially HDL-C (p=0.009 for total cholesterol and HDL-C, and p=0.03 for LDL-C). Women with sPL had increased AI1 (total cholesterol/HDL-C).

Proinflammatory cytokines and markers of endothelial dysfunction are shown in Table 2. Maternal serum levels of IL-6 were higher and MPO levels were lower in women with sPL (p=0.007 and p=0.0009 respectively) than in controls.

Correlations between variables having significant differences between control and sPL women are shown in Table 3. In the case of smoking, nonsignificant values could be a result of the small sample of smokers in each group. Because of the absence of significant relationships, we decided not to include this variable in the correlation and in the multivariate predictive model. All lipid fractions correlated with each other. Glycemia was significantly correlated with total cholesterol (p=0.03). Maternal serum IL-6 was significantly correlated with both glycemia (p=0.02) and the AI1 (p=0.04), and it had borderline correlations with HDL-C. Maternal MPO levels were strongly correlated with BMI (p<0.0001) and had borderline correlations with glycemia and LDL-C levels. Forward stepwise binary logistic regression showed that the best predictive model for sPL (predicted 69.8% correct, R ² =0.36, p=0.001) included only two variables: BMI (odds ratio [OR] 0.74, confidence interval [95% CI 0.55-0.99]) and total cholesterol (OR 0.97, 95% CI 0.45-0.99).

Discussion

Our aim was to identify metabolic CVD risk factors in women with sPL and to elucidate whether or not they were related to inflammation or endothelial damage. We found that total cholesterol and lipoproteins levels were decreased in women with sPL. Proportionally, the most significant reduction was found in HDL-C concentrations. Consequently, despite having low total cholesterol levels, the AI (total cholesterol/HDL-C) was significantly higher in women with sPL. This ratio was correlated with IL-6, and we speculate that these finding may be related to an increased cardiovascular risk later in life in these women.

There are only a limited number of studies focusing on the association between dyslipidemia and sPL. Alterations in maternal cholesterol levels have been related to increased rates of sPL. Both high ^10,12 and low ¹³ maternal cholesterol levels have been related to a higher incidence of this condition. Catov et al. ¹⁰ found that high cholesterol or triglyceride levels in early pregnancy were associated with an increased risk for preterm birth. There are several differences in our study. Cases of PROM were included, lipid levels were evaluated in early pregnancy, and both hypertriglyceridemia and hypercholesterolemia were pooled in one combined outcome variable.

Along this line, another study ¹⁴ demonstrated that a cholesterol-lowering diet reduced dramatically the rate of preterm delivery in low-risk pregnancies. In our opinion, the selection criteria for this sample could have influenced their impressive result (only 0.7% of prematurity in the diet group), and it is unlikely it can be extrapolated to the general population. In another study, it was reported that the effect of this cholesterol-lowering diet was not related to changes in the systemic inflammatory responses of pregnancy. ¹⁵

In opposition and consistent with our results, it has been reported that pregnant women with low total cholesterol levels had an increased rate of sPL (12.7%) compared with control subjects (5.0%). ¹³ Thus, the authors suggested caution when using a cholesterol-lowering diet if cholesterol levels fell below the normal range. Finally, Catov et al. ¹⁶ reported that both high and low cholesterol levels can be related to preterm birth, perhaps representing distinct pathways to the heterogeneous outcome of preterm birth; however, but that study used prepregnancy lipid measurements.

Previous studies have reported increased maternal levels of proinflammatory markers in women with sPL, most including patients with PROM; this inflammatory component has been classically related to the presence of genital, fetoplacental, or systemic infection. ^17,18 After excluding these women, the results relating to maternal levels of circulating cytokines have been inconsistent. ^{19 –22} We found significantly increased levels of IL-6, but this variable was significantly correlated with both the total cholesterol/HDL-C ratio and glycemia levels. Hyperglycemia is a known risk factor for both CVD and sPL, and both conditions also are related to inflammation. In our study, however, no cases of diabetes were included, and glucose levels were not significantly higher in the sPL group. Therefore, although glycemia was related to IL-6 levels, it seems unlikely that this metabolic component is related to the presence of sPL. Further studies are needed to clarify the role of inflammation specifically in women with diabetes mellitus and sPL.

Inflammation has been associated with both dyslipidemia and the metabolic syndrome. Previous publications have reported that when all the metabolic syndrome components are entered simultaneously into one analysis of covariance (ANCOVA) model, only low HDL-C levels proved to be associated with C-reactive protein. ²³ These data fit into the notion that HDL particles, besides their crucial role in reverse cholesterol transport, also protect the artery wall through anti-inflammatory mechanisms. HDL may have an inhibitory effect on apoptosis, lipid oxidation, cytokine, and adhesion molecule production, and its stimulatory effects on endothelial function support the relationship between low HDL-C levels and inflammation and cardiovascular risk. ²⁴

On the other hand, it is not completely clear if both low HDL-C and total cholesterol levels could be a consequence rather than a cause of inflammation. It has been demonstrated that both acute-phase reactions and chronic inflammation reduce the concentration of total cholesterol, LDL-C, and HDL-C and alter the apolipoprotein and lipid composition of HDL. This can be seen in sepsis, where HDL-C is reduced by 50%, ¹¹ in acute infections in childhood, ²⁵ and in chronic inflammation, such as periodontitis and rheumatoid arthritis. ²⁶ Consequently, although we have found a significant correlation between IL-6 and the ratio of total cholesterol/HDL-C, we cannot determine if inflammation was a cause or consequence of low HDL-C levels. If it was a consequence, this may explain why other authors have found normal HDL-C levels in early pregnancy in women who subsequently develop sPL because in early pregnancy, it is supposed that the inflammatory condition it is not present.

We have not found any difference between women with sPL and controls in terms of endothelial dysfunction markers. In a previous study ²⁶ on women with sPL, including those with PPROM, increased maternal serum levels of ICAM were found in women with histologic chorioamnionitis, but normal levels were found in those without histologic chorioamnionitis. The authors determined that ICAM-1 concentrations are influenced only by infective processes and not by preterm labor.

We found decreased MPO levels in the sPL group. Apart from its role in atherogenesis, MPO is a microbicidal hemoprotein that serves as part of the neutrophil activation in host defense. In normal women, MPO levels decrease during pregnancy, although a deeper decrease in MPO level may be related to a higher risk for infection. It has been shown that a reduction in the level of MPO in placentas of women with sPL could favor infection of the fetus/placenta unit. ²⁷ Interestingly, we found MPO levels related to BMI. Some previous studies ²⁸ have found a relationship between vaginal inflammation and low maternal BMI. Whereas obesity is related to induced preterm labor (mainly as a consequence of other associated pregnancy complications, such as diabetes mellitus or hypertension), maternal underweight is a known risk factor for sPL. ^{29 –31} We found that women with sPL weighted less than controls. Even binary logistic regression showed that BMI was one of the best predictors for preterm delivery, having a protective role. The reasons for this association remain unclear. It could be related to the presence of lipid levels, but it could be also related to other risk factors for sPL, such as infection. Further studies are needed to clarify the role of MPO in the genesis of sPL.

In conclusion, low BMI and total cholesterol levels combined with a high total cholesterol/HDL-C ratio were present in women with sPL. This increased AI was associated with inflammation. We speculate that the long-term association between sPL and CVD could suggest that an upregulation of chronic inflammatory pathways may be involved in the pathogenesis of both diseases. The increased cardiovascular and metabolic physiologic demands that occur during pregnancy may unmask subclinical conditions that may disappear after pregnancy and emerge later in a woman's life. ^32,33 Women with a proinflammatory phenotype may develop greater upregulation of the chronic inflammatory pathways than that seen in normal pregnancy, leading to both sPL and CVD in later life, and these conditions may be associated with a specific lipid pattern. Confirmation of this hypothesis needs further investigation.