Cigarette Smoke Exposure and Angiogenic Factors in Pregnancy and Preeclampsia

Abstract

Background

Cigarette smoking during pregnancy is paradoxically associated with a reduced risk of developing preeclampsia. Both smoking and preeclampsia are associated with alterations in circulating angiogenic factors. The objective of this study was to investigate the relationship between cigarette smoking and the angiogenic factors soluble fms-like tyrosine kinase-1 (sFlt-1) and placental growth factor (PlGF) in pregnant women with and without preeclampsia.

Methods

Plasma sFlt-1, PlGF and cotinine were measured by ELISA in 125 women with uncomplicated pregnancies (controls) and 58 women with preeclampsia.

Results

In uncomplicated pregnancies, maternal sFlt-1 concentrations were lower in smokers compared to nonsmokers (779.6 [487.5-1140.8] vs. 1116.5 [793.6-1905.2] pg/ml, p<0.005). Preeclamptic women who smoked also demonstrated a trend towards lower concentrations of sFlt-1 compared to nonsmokers (3423.0 [2183.4-5689.0] vs. 5504.9 [3418.0-6361.3] pg/ml, p=0.07). Maternal PlGF concentrations were higher in smokers with uncomplicated pregnancies (398.4 [165.2-621.7] vs. 191.4 [104.6-446.8] pg/ml); however, this was not a statistically significant difference (p=0.07). PlGF concentrations were not different in preeclamptic smokers compared to nonsmokers. The sFlt/PlGF ratio was significantly lower in smokers with uncomplicated pregnancies, but not in smokers with preeclampsia compared to nonsmokers.

Conclusions

Cigarette smoking is associated with lower maternal sFlt-1 concentrations during pregnancy and preeclampsia. Based on these data, cigarette smoke exposure may decrease the risk of preeclampsia in part by moderating the anti-angiogenic phenotype observed in the syndrome.

Introduction

Cigarette smoke exposure is associated with many adverse pregnancy effects including preterm labor, preterm premature rupture of membranes, placental abruption and fetal growth restriction/low birth weight.¹ Paradoxically, smoking is associated with a decreased risk of preeclampsia.^2-4 Mechanism(s) by which cigarette smoke exposure exerts these effects on pregnancy, including the risk reduction of preeclampsia, is unclear.

The role of angiogenic growth factors, particularly placental growth factor (PlGF), vascular endothelial growth factor (VEGF) and the anti-angiogenic soluble VEGF receptor 1 (soluble fms-like tyrosine kinase-1 or sFlt-1) in the pathogenesis of preeclampsia has received increased attention over the past few years. Angiogenic growth factors likely play important roles in maternal vascular function and remodeling of the maternal vasculature during pregnancy including uterine spiral arteries perfusing the placenta. Soluble Flt-1 binds VEGF and PlGF, competes with the binding of receptors located on the endothelium, and is associated with impaired vascular function.^5-7 Levine et al. have reported significant differences in these angiogenic factors predating the clinical manifestations of preeclampsia.⁸ Specifically, lower concentrations of PlGF have been noted, while sFlt-1 concentrations are higher. Cigarette smoking may also influence these angiogenic factors with lower circulating sFlt-1 concentrations in nonpregnant smokers compared to nonsmokers.^{9, 10} Recent evidence suggests that cigarette smoking in pregnant women may also lower maternal sFlt-1 concentrations.^11-13 There is limited information on the effect of cigarette smoke exposure on maternal PlGF concentrations.^{13, 14}

The objectives of this study were to investigate the association of circulating sFlt-1 and PlGF concentrations with maternal smoking in women with and without preeclampsia. We hypothesized that cigarette smoking would be associated with lower concentrations of sFlt-1 and higher concentrations of PlGF in both uncomplicated pregnant women and women with preeclampsia.

Methods

Study population

This was a nested case-control study of 183 women enrolled in an ongoing investigation of preeclampsia at Magee-Womens Hospital (Pittsburgh, PA). The study was approved by the Institutional Review Board and informed consent was obtained from all subjects. Fifty eight (58) women had preeclampsia (cases) and 125 women had uncomplicated pregnancies (controls). Subjects were also selected based on self-report of smoking with 50% smokers and 50% nonsmokers in each group. Samples were matched for age, race, body mass index, and gestational age at the time of sample collection (Table 1). Exclusion criteria included multiple gestation, prior preeclampsia, illicit drug use and preexisting medical conditions such as diabetes, chronic hypertension, and renal disease. Preeclampsia was diagnosed by the presence of gestational hypertension (an absolute blood pressure ≥ 140 mm Hg systolic and/or 90 mm Hg diastolic after 20 weeks of gestation), proteinuria (greater than 300 mg per 24-hour urine collection, ≥2+ on a voided or ≥1+ on a catheterized random urine sample, or a protein/creatinine ratio of >0.3), and hyperuricemia (≥ 1 standard deviation above reference values for the gestational age the sample was obtained (e.g. term, > 5.5 mg/dL)) beginning after the 20^th week of pregnancy with resolution of blood pressure and proteinuria postpartum. We include hyperuricemia in our classification as it identifies a more homogeneous group of gestational hypertensive women with a greater frequency of adverse fetal outcomes.¹⁵ In this study, we determined blood pressure as the average of the last 5 blood pressures obtained in semi-recumbent position after hospital admission but before medications or clinical perturbations that would alter blood pressure. Diagnosis of preeclampsia was determined retrospectively based on medical chart review by a jury of research and clinical investigators. Smoking status was determined by self-report and confirmed by plasma cotinine measurements. The self-report of smoking was obtained at the time of sample collection or immediately postpartum to minimize the potential problems caused by changing smoking habits during pregnancy. Plasma cotinine concentrations of ≥ 2.0 ng/ml were chosen as indicative of smoke exposure, while concentrations below this cutoff indicated nonsmoking individuals.^{16, 17}

Table 1.

Maternal and Newborn Demographics

	Control pregnancy/non-smoker (n=64)	Control pregnancy/smoker (n=61)	Preeclampsia/non-smoker (n=30)	Preeclampsia/smoker (n=28)
Maternal age (years)	22.8±5.4	22.6±4.3	27.7±5.6*	22.0±3.8
Maternal prepregnancy body mass index (kg/m2)	25.6±6.4	24.2±6.6	25.8±8.9	25.9±6.0
Average blood pressure before 20 weeks gestation (mmHg)	112.7±7.8/69.0±6.3	111.2±6.4/67.0±5.3	115.6±10.8/72.1±7.6	115.7±9.9/70.1±8.2
Average blood pressure at delivery (mmHg)	119.7±11.3/71.9±6.5	117.5±10.8/70.4±7.5	157.3±15.8*/94.4±8.8*	157.2±19.9*/91.8±12.5*
Race (% African American)	37%	21%	21%	21%
Maternal plasma uric acid at delivery (mg/dl)	4.4±1.0	4.9±1.1	6.4±1.3*	6.8±1.7*
Maternal plasma cotinine (ng/ml)	1.6±3.8	61.3±56.9*	0.8±1.4	30.1±43.5*
Gestational age at delivery (weeks)	37.5±2.3	37.9±2.3	34.8±3.6*	34.8±3.9*
Gestational age at sample collection (weeks)	34.3±3.8	34.1±4.4	34.6±3.6	34.1±4.4
Infant birth weight (grams)	2946±525	3022±616	2125±811*	2045±827*
Infant birth weight centile (%)†	50.9±24.4	44.7±25.9	29.5±26.4*	21.9±22.0*

Data are mean±SD,

^*p<0.05 compared to control pregnancy/ non-smoker.

^†Infant birth weight centile derived from nomograms based on race, gender, and gestational age from a reference population of over 10,000 neonates delivered at Magee-Womens Hospital.

Blood samples

Maternal venous EDTA-treated plasma samples were collected during the third trimester of pregnancy and stored at -70°C until assayed. Gestational age was determined by best obstetric estimate (first trimester ultrasound when available). By design the gestational age at sample collection was not significantly different between the controls and women with preeclampsia (Table 1).

Quantitation of sFlt-1 and PlGF

Plasma concentrations of sFlt-1 and PlGF were measured using commercial ELISA kits from R&D systems (Minneapolis, MN) according to the manufacturer's protocol. The sensitivity for PlGF was 7 pg/ml and the assay specific for PlGF without VEGF cross reactivity. The sFlt assay is specific for human sFlt-1, with sensitivity of 5 pg/ml. Inter-assay variability for sFlt-1 and PlGF was 12% and 7%, respectively.

Plasma cotinine measurement

Plasma cotinine was measured using the commercially available ELISA from Immunalysis Corp (Pomona, CA) according to the manufacturer's protocol with an inter-assay variation of <10% and sensitivity of 1 ng/ml.

Statistical analysis

Based on our previous published work ¹¹, sample size calculations indicated that at least 24 subjects per group with preeclampsia and 57 subjects per group with uncomplicated pregnancies would be required to observe an effect size due to smoking of 0.5 and 0.6, respectively, with 80% power and an alpha of 0.05. Data are presented as mean ± standard deviation or median and inter-quartile ranges where appropriate. Plasma concentrations of sFlt-1 and PlGF were not normally distributed, therefore log transformed values, which normalized the data distribution, were used for the statistical analyses. JMP 5.0.1a (SAS Institute, Cary, NC) and Stata 9 (StataCorp, College Station, TX) software were used to analyze the data. Analysis of variance was performed with Fisher's PLSD post hoc testing to adjust for multiple comparisons. Regression analysis was used to determine the association of smoking to sFlt-1 and PlGF among controls and preeclamptic women. Wilcoxon rank sum test was used to compare the ratio of sFlt-1/PlGF due to the nonparametric distribution. A p-value of <0.05 was considered statistically significant. Self-report of smoking and plasma cotinine were compared using McNemar's test and degree of agreement described using the kappa coefficient.

Results

The demographic and clinical characteristics of the four groups: smoking and nonsmoking, preeclamptic and control subjects are presented in Table 1. The mean gestational age at time of sample collection was not different between groups. There were also no significant differences in race, body mass index, or blood pressure prior to 20 weeks' gestation. As expected, the gestational age at delivery was significantly lower in the preeclamptic women compared to control subjects. Maternal blood pressure at delivery and plasma uric acid were higher, while infant birth weight and birth weight percentile were lower in both smokers and nonsmokers with preeclampsia compared to controls.

Maternal cigarette smoke exposure was primarily assessed by self-report of smoking. Mean cotinine concentrations were significantly higher in the smoking group compared to the nonsmoking group (p<0.05 for both the preeclamptic and control women, Table 1). There was no significant discordance between self-report of smoking status and maternal plasma cotinine concentration using a cut-off value of ≥2.0 ng/ml (p = 0.49), with 82% agreement and kappa coefficient of 0.64. Because of the varying time intervals between the last cigarette smoke exposure and the blood draw, particularly with preeclamptic women who were often hospitalized prior to the time of the blood draw, we chose the subject's self-report to represent smoking status for our primary analysis of sFlt-1 and PlGF.

Among women with uncomplicated pregnancies (controls), maternal plasma sFlt-1 concentrations were lower in smokers compared to non-smokers, 779.6 [487.5-1140.8] vs. 1116.5 [793.6-1905.2] pg/ml, p<0.005 (Figure 1A). In women with preeclampsia, plasma sFlt-1 concentrations were lower in smokers compared to nonsmokers, 3423.0 [2183.4-5689.0] vs. 5504.9 [3418.0-6361.3] pg/ml, by marginal significance (p=0.07) (Figure 1B). Among control subjects, plasma PlGF concentrations were higher in smokers compared to control subjects who were not smokers, 398.4 [165.2-621.7] vs. 191.4 [104.6-446.8] pg/ml (Figure 2A); however, this was not a statistically significant difference (p=0.07). There was no difference in maternal PlGF concentrations between smokers and nonsmokers with preeclampsia, 28.7 [16.1-67.9] vs. 28.7 [13.2-53.7] pg/ml, p=0.44 (Figure 2B). When the data were analyzed by cotinine concentration using cut-offs of 2, 5, or 10 ng/ml, there was no remarkable change in the significance of sFlt1 and PlGF differences between smokers and nonsmokers compared to data obtained by self-report (data not shown). We also evaluated the ratio of sFlt-1 to PlGF which may be more representative of the overall angiogenic milieu. In women with uncomplicated pregnancies, smokers had a lower sFlt-1/PlGF ratio (1.64 [0.94-10.0] vs 4.85 [2.18-13.03], p=0.005). Similarly, in women with preeclampsia the sFlt-1/PlGF ratio was lower for smokers compared to nonsmokers (89.42 [54.41-310.51] vs 158.41 [94.39-346.79]; albeit not statistically significant (p=0.18).

Figure 1. Box and whisker plots of maternal plasma sFlt-1 in women who smoked during pregnancy and nonsmokers with uncomplicated pregnancies (controls) and pregnancies complicated by preeclampsia.

The filled black circles are the median, the open circles are the 90th and 10th percentiles, and the top and bottom lines of the box are the 75^th and 25^th percentiles of the data for each group.

Figure 2. Box and whisker plots of maternal plasma PlGF in women who smoked during pregnancy and nonsmokers with uncomplicated pregnancies (controls) and pregnancies complicated by preeclampsia.

The filled black circles are the median, the open circles are the 90th and 10th percentiles, and the top and bottom lines of the box are the 75^th and 25^th percentiles of the data for each group.

In addition to the univariate analyses above, we performed regression analyses for sFlt1 and PLGF using the entire study population and evaluated the main effects of pregnancy outcome, smoking status, as well as their interaction. Additional co-variates, maternal age, race, BMI, and gestational age at the time of sample collection were included in the regression model. As expected, gestational age at sampling was significantly associated with sFlt-1 and PLGF. Importantly, this analysis demonstrated an independent effect of smoking on sFlt-1 (p=0.001) and a trend for its effect on PlGF (p=0.08) with no apparent interaction between pregnancy outcome and smoking status on either sFlt-1 or PlGF (p=0.67 and 0.45, respectively).

Comment

The overall objective of our study was to further elucidate the effect of cigarette smoking on circulating angiogenic factors in women with and without preeclampsia. This work builds on the findings of lower sFlt-1 concentrations in nonpregnant smokers ^{9, 10}, recent work indicating a pathophysiologic role of angiogenic factors in preeclampsia ^{8, 18}, and the paradoxical association of smoking and a reduced risk of preeclampsia. ^{2, 3} Our primary results indicate that smoking is associated with lower sFlt-1 concentrations regardless of pregnancy outcome, although this was less apparent in women with established preeclampsia. Smokers with uncomplicated pregnancies have higher concentrations of PlGF, although this elevation was not statistically significant. This difference was not apparent in preeclamptic pregnancies.

While the association of preeclampsia with higher sFlt-1 and lower PlGF concentrations in the maternal circulation has been extensively reported, the effect of smoking on the angiogenic milieu of pregnancy is less well defined. Our observation that smoking is associated with lower sFlt-1 concentrations in the maternal circulation supports much of the existing data: First, cigarette smoking is associated with lower sFlt-1 levels in healthy, nonpregnant subjects.^{9, 10} Second, Levine and colleagues demonstrated lower levels of sFlt-1 in smokers with uncomplicated pregnancies in a cohort of subjects from the Calcium for Preeclampsia Prevention trial, but these data were not confirmed by cotinine measurements.^{12, 13} Third, our group has previously reported that sFlt-1 concentrations were significantly lower in smokers with uncomplicated pregnancies at 15 to 34.5 weeks', 35 weeks to term, and even postpartum.¹¹ However, these conclusions were limited by the sample size and lack of cotinine confirmation of smoking self-report. Importantly, the relationship of smoking to sFlt-1 concentrations during and prior to preeclampsia has not been well described. In our prior study, smoking did not significantly affect sFlt-1 levels at delivery in women with preeclampsia or prior to the development of preeclampsia; however, conclusions were limited by the small number of preeclamptic women.¹¹ In contrast to sFlt-1, published information on the effect of cigarette smoking on PlGF is limited,^{13, 14} and to our knowledge there are no published studies to date evaluating the effect of cigarette smoke exposure on circulating PlGF in preeclamptic women. We observed a possible trend towards higher concentrations of circulating PlGF with smoking, however, this effect was not statistically significant in preeclamptic and uncomplicated pregnancies. This could be an issue of power as our sample size was calculated based on known differences in sFlt-1; furthermore, PlGF concentrations in our study were highly variable. Here it is worthwhile to note that our study adds to the existing literature, because (1) we use a different cohort of subjects which is particularly important since the effects of smoking on PlGF have been tested in only one cohort^12,13 as published to date, (2) we evaluate the relationship of sFlt-1 and PlGF in both control and preeclamptic subjects, and (3) our conclusions are strengthened by our sample size and confirmation of self-reported smoking status by cotinine measurements.

While the precise mechanism(s) of smoking's effect on these angiogenic factors is unknown, recent work by Mehendale and colleagues demonstrated that exposure of placental villous explants to cigarette smoke extract, reduced sFlt-1 concentrations in the media.¹⁹ Additionally, the observation that cigarette smoke increases heme oxygenase expression in the basal plate of the placenta²⁰ and the inhibitory effects of heme oxygenase-1 and carbon monoxide on sFlt-1 release²¹ provides a possible biologic mechanism by which smoking may lower sFlt-1 concentrations in pregnancy. Interestingly, our data on PlGF concentrations in smokers are similar to the nonsignificant trend towards higher PlGF concentrations at 48 and 72 hours in the media of placental villous explants exposed to cigarette smoke extract.¹⁹ These in vitro findings are in line with our observations that the association between smoking and sFlt-1 is robust, whereas the association with PlGF is less clear. Additional in vivo and in vitro investigations are warranted to fully establish the effect of cigarette smoke exposure on PlGF.

The ratio of sFlt-1/PlGF, which is currently used by several investigators to represent the overall angiogenic milieu, was altered in the same direction as sFlt-1 in smokers compared to nonsmokers in both uncomplicated and preeclamptic pregnancies. This alteration suggests a shift towards a pro-angiogenic phenotype in the maternal circulation with smoking. We also analyzed our data to determine whether smoking exerted a differential effect on these angiogenic factors in women with preeclampsia and uncomplicated subjects. Our results suggest that there was no apparent interaction between preeclampsia and smoking in our regression model suggesting that smoking exerts a similar effect on sFlt-1 and PlGF based on pregnancy outcome (preeclampsia and uncomplicated pregnancy). The absence of an interaction does not exclude the possibility of risk reduction via angiogenic pathways. Importantly, our univariate analyses suggest important modulatory effects of smoking on angiogenic factors; however, it is important to consider that these samples were drawn after the established diagnosis of preeclampsia. While the marked alterations of these factors after clinically apparent disease must be considered in the interpretation of our results, overall our results support that the risk reduction may be via modulation of angiogenic pathways.

We chose to use self-report of smoking as our primary method of classification in our study. Misclassification of maternal smoking status is a recognized problem in pregnancy research.²² On the other hand, the use of cotinine alone may not precisely represent smoking status in our subjects. Many of the blood samples, particularly in preeclamptic women, were drawn after hospitalization and at varying intervals from the last cigarette smoke. The reduced half-life of cotinine in pregnancy (8.8 hours in pregnancy compared to 16.6 hours in nonpregnant subjects)²³ along with the unknown time course of smoking's effect on angiogenic factors could affect the interpretation of the results. Therefore to avoid misclassification bias, we used self-report in conjunction with plasma cotinine in our analysis. It is important to note that we observed 82% agreement between smoking self-report and cotinine concentration and no difference in the observed trends when the data were analyzed by cotinine concentration instead of smoking self-report.

There is an abundance of data that has consistently shown that cigarette smoking during pregnancy is associated with adverse pregnancy outcomes; paradoxically despite this, the risk of the pregnancy complication preeclampsia is lower among smokers compared to nonsmokers. There is little information regarding potential biological mechanisms that may help explain this paradox. The data presented in this manuscript suggests one such mechanism, namely that cigarette smoke exposure may modulate the angiogenic milieu of pregnancy by affecting concentrations of sFlt-1 and possibly, PlGF. However, these effects require additional investigation, as there are limitations of the present study including its case-control and retrospective design; specifically, blood was drawn in the third trimester of pregnancy and after the clinical diagnosis of preeclampsia was established. With preeclampsia, the markedly higher or lower sFlt1 and PlGF respectively, may dampen or obscure the effects of smoking on these circulating factors. Secondly, our sample size calculations were based on existing information of sFlt-1 in pregnancy and preeclampsia, which may limit conclusions regarding the association of PlGF and smoking in preeclamptic and uncomplicated pregnancies. Another important consideration is that our study is limited to maternal circulating concentrations and therefore does not address the question of the precise tissue source of these angiogenic factors (placenta, endothelium, or peripheral mononuclear cells) nor does it address the mechanisms by which smoking affects circulating concentrations of these factors. Despite these limitations, our findings indicate a modulatory effect of smoking on angiogenic factors during pregnancy and suggest a possible mechanism by which smoking may reduce the risk of preeclampsia.