Objective
The purpose of this study was to describe relations among maternal demographic and lifestyle characteristics and midpregnancy levels of angiogenic markers (soluble Fms-like tyrosine kinase-1, placental growth factor, soluble endoglin).
Study Design
In a large pregnancy cohort, linear models were used to evaluate relations among maternal characteristics and midpregnancy angiogenic markers with and without covariate adjustment. Associations were examined in a subcohort that included term and preterm deliveries (n = 1302) and among “normal” term pregnancies (n = 668).
Results
Concentrations of all factors declined with increasing maternal body mass index. Multiparous women had lower soluble Fms-like tyrosine kinase-1 levels than primiparous women. Higher placental growth factor and slightly lower soluble endoglin levels were observed among women who smoked at enrollment, but not among those women who quit before enrollment. African American women had higher levels of all markers.
Conclusion
Understanding relations among maternal characteristics and levels of angiogenic factors may improve studies that use these markers to examine etiology and/or to predict adverse pregnancy outcome.
Altered patterns of angiogenic markers have been linked to pregnancy conditions that are responsible for significant maternal and fetal morbidity. These observations may offer etiologic insights and translate into the use of the markers for disease screening. For preeclampsia, which is the condition with the most extensive and strongest evidence, elevated concentrations of soluble Fms-like tyrosine kinase 1 (sFlt-1) and soluble endoglin (sEng) and lower levels of placental growth factor (PlGF) have been detected months before disease onset. Additionally, several altered patterns of angiogenic markers may contribute to poor fetal growth, hypertension-related placental abruption, spontaneous preterm birth, proteinuria, gestational hypertension, and fetal death.
During pregnancy sFlt-1, PlGF, and sEng are produced by placental tissue, although decidual cells, leukocytes, and other cells may contribute to overall levels. Measured in maternal serum, levels of sFlt1 and sEng are fairly steady early in pregnancy and rise more steeply later, whereas PlGF levels rise steadily into the third trimester and decline the last 8-10 weeks of pregnancy. Angiogenic factors, either working independently or by modulating binding of vascular endothelial growth factor and transforming growth factor–β to their receptors, may regulate placental angiogenesis. Additionally, these factors affect the maternal endothelium and, in cases of aberrant expression, lead to the development of symptoms such as gestational hypertension and proteinuria. Although this is linked to placental problems, it is unclear whether altered expression is the cause or consequence of poor placental development and function.
Some studies have examined relations among these angiogenic markers and maternal characteristics, but many have been small, have examined only 1 or 2 characteristics or markers, and have lacked adjustment for other covariates. In addition, studies have varied in their inclusion/exclusion of complicated pregnancies. A more complete understanding of associations between angiogenic markers and maternal characteristics may provide clues to the underlying mechanism and may improve control of confounding in etiologic studies or those studies testing the screening/diagnostic potential of these markers.
In the Pregnancy Outcomes and Community Health Study (POUCH), we describe associations among maternal demographic, anthropometric, and lifestyle characteristics and each of 3 angiogenic markers that were measured in maternal serum at midpregnancy. Associations are examined within a group of “normal” term pregnancies in which women did not experience the most prevalent complications that are associated with altered angiogenic patterns; this permitted a clearer interpretation of relations. For comparison, we also examined associations among unselected pregnancies that represented a spectrum of pregnancy outcomes, which may indicate the degree to which processes underlying poor pregnancy outcomes influence findings.
Materials and Methods
Population
The POUCH Study recruited pregnant women from 52 prenatal clinics in 5 different Michigan communities between September 1998 and June 2004. Eligible women had maternal serum α-fetoprotein (MSAFP) screening at 15-22 weeks’ gestation, had singleton pregnancies with no known abnormality or birth defect, were ≥15 years of age, were English-speaking, and had no diagnosis of prepregnancy diabetes mellitus. Given the study’s focus on preterm birth and previous reports that linked elevated MSAFP to preterm delivery, all women with unexplained, elevated MSAFP levels (≥2.0 multiples of the median) were invited to participate, which enriched the sample for women with elevated levels (7% vs 3.5% in general obstetric population). A total of 3038 women were enrolled between 15 and 27 weeks gestation, at which time self-completed questionnaires and in-person interviews were administered and maternal biologic samples were collected. In total, 3019 women (99.5%) who were followed through delivery comprise the POUCH cohort. The POUCH Study protocol was approved by the institutional review boards at Michigan State University, the Michigan Department of Community Health, and all delivery hospitals.
Study sample
To use resources in a way that maximized statistical power for studying preterm delivery and at-risk groups such as women with elevated MSAFP levels and African American women, we established a subcohort of women (n = 1371) in which more detailed information was obtained. The subcohort included all women who delivered preterm, all women with elevated MSAFP levels, and a sample of term births with normal MSAFP levels and oversampling of African American women. To account for oversampling of preterm deliveries and at-risk groups, analyses were weighted according to the probabilities of selection into the cohort and subcohort. (For example, the proportion of individuals with high MSAFP levels in the entire cohort was approximately 7%, which is twice that of the general population; therefore, this group would be assigned a cohort adjustment weight of one-half.) Additional weights would be calculated for sampling into the subcohort. The weighted results should reflect the experience of the population that was sampled, as previously demonstrated for demographic characteristics and selected pregnancy outcomes.
Women who were included in this analysis belonged to the subcohort, had measures of angiogenic factors available, and were not missing information on important covariates (n = 1302; Figure ). Additionally, we conducted a subset analysis with women who had “normal” pregnancies, which were defined as (1) term delivery (>37 weeks’ gestational age); (2) MSAFP <2.0 multiples of the median; (3) delivery of a child who was not small-for-gestational age, which was defined <10th percentile of weight for gestational age); and (4) no diagnosis of any hypertensive disorder. Hypertensive disorders that were ascertained through review of prenatal and labor/delivery records by trained abstractors included diagnoses of chronic hypertension before pregnancy, gestational hypertension (minimal criteria: diastolic blood pressure ≥90 or systolic blood pressure ≥140 on 2 different days after 20 weeks gestation without evidence of proteinuria), or preeclampsia (minimal criteria: gestational hypertension plus proteinuria that was defined as 2+ protein on urine dipstick once or 1+ protein on 2 occasions after 20 weeks’ gestation). If proteinuria was present at <20 weeks, increased levels at >20 weeks were considered indicative of preeclampsia.
Covariates
Maternal demographic, reproductive, lifestyle, and anthropometric characteristics were obtained by interview or questionnaire that was administered at enrollment. Race was categorized for these analyses as African American or non-African American. Medicaid enrollment (yes/no) and completed years of education were proxies for socioeconomic status. Smoking was grouped into 4-levels: did not smoke during pregnancy, smoked in pregnancy but stopped before enrollment, smoked less than one-half pack per day at enrollment, and smoked one-half pack or more per day at enrollment. Prepregnancy weight and maternal height were obtained from the questionnaire. These 2 measures were used to calculate prepregnancy body mass index (BMI), which was grouped according to Centers for Disease Control guidelines into underweight, normal, overweight, and obese. Gestational age at study enrollment, which was grouped as 15 to <20, 20 to <25, and 25 to <28 weeks, was based on last menstrual period if this estimate was within 2 weeks of the date that had been estimated from a <25-week ultrasound examination. Otherwise ultrasound dates were used.
Measurement of angiogenic markers
Maternal serum samples that were collected at enrollment were stored at –80°C. For analysis, samples were allocated randomly to batches by study site and length of postcollection storage time. Additionally, all batches contained the same proportion of preterm deliveries because this outcome was a primary focus of the study. Measurements of sFlt-1, PlGF, and sEng were made with commercially available enzyme-linked immunosorbent assay kits (R&D Biosystems, Minneapolis, MN) by 1 Karumanchi Laboratory member who was blinded to patients’ clinical data. All measurements were performed in duplicate. Assays were rerun if measures differed by >25%. Limits of detection for each marker were 7 pg/mL (sFlt-1), 5 ng/mL (PlGF), and 5 pg/mL (sEng). Interassay coefficients of variation were 7.6%, 10.9%, and 3.0% for sFlt-1, PlGF, and sEng, respectively. Intraassay coefficients were 3.3%, 5.6%, and 6.3%, respectively. Duplicate measures were averaged and log-transformed to create approximately normal distributions for use in general linear models.
Analysis
Associations between angiogenic markers and maternal characteristics were examined in 2 groups of women: all POUCH pregnancies with complete biomarker and covariate measurements (95% of subcohort women; n = 1302) and a sample of “normal” pregnancies (term, normal MSAFP levels, no hypertensive disorders, not small-for-gestational age; n = 668). Analyses were completed in SAS software (version 9.1.3 or 9.2; SAS Institute Inc, Cary, NC) and were weighted appropriately with the use of Survey Procedures. Frequency distributions were calculated with Proc SurveyFreq (SAS Institute Inc). Linear regression was implemented in Proc SurveyReg (SAS Institute Inc) to determine the regression coefficient for each angiogenic marker regressed on covariates of interest. Covariates were examined individually and in multivariable models. Unadjusted and adjusted least squares means and 95% CIs are presented for each factor at different levels of maternal demographic, lifestyle, and anthropometric variables.
Results
The weighted frequencies showed little difference between the distributions of maternal characteristics with the entire POUCH subcohort and women with complete measurements ( Table 1 ). Approximately 24% of women were African American; slightly more than one-half of the women were 20-29 years of age; 41% of the women were primiparous; >25% of the women reported having smoked at some time during pregnancy, and most of the women enrolled between 20 and 24 weeks’ gestation at a mean 22.4 and 22.5 gestational weeks for the total and “normal” pregnancy samples, respectively ( Table 1 ).
| Characteristic | Entire subcohort (n = 1371) | Total pregnancies, sample (n = 1302) | Normal pregnancies, sample (n = 668) | |||
|---|---|---|---|---|---|---|
| n | Weighted, (%) | n | Weighted, (%) | n | Weighted, (%) | |
| Maternal | ||||||
| Race | ||||||
| White | 692 | 65.8 | 668 | 66.7 | 316 | 66.8 |
| African American | 579 | 24.6 | 536 | 24.6 | 305 | 23.3 |
| Asian | 23 | 2.2 | 22 | 2.1 | 9 | 1.9 |
| Hispanic | 58 | 5.1 | 58 | 5.3 | 26 | 5.5 |
| Other | 19 | 2.3 | 18 | 2.4 | 12 | 2.5 |
| Age, y | ||||||
| <20 | 243 | 14.7 | 227 | 14.6 | 113 | 13.2 |
| 20-29 | 776 | 57.2 | 734 | 56.8 | 401 | 59.0 |
| ≥30 | 352 | 28.1 | 341 | 28.6 | 154 | 27.8 |
| Parity | ||||||
| Primiparous | 577 | 41.5 | 543 | 41.0 | 256 | 38.3 |
| Multiparous | 793 | 58.5 | 759 | 59.0 | 412 | 61.7 |
| Missing | 1 | 0 | — | — | — | — |
| Education: highest level attained, grade | ||||||
| <12th | 317 | 18.9 | 296 | 18.8 | 154 | 17.7 |
| 12th | 388 | 27.0 | 367 | 27.1 | 184 | 26.1 |
| >12th | 666 | 54.1 | 639 | 54.1 | 330 | 56.2 |
| Medicaid | ||||||
| No | 586 | 50.9 | 565 | 50.8 | 286 | 52.6 |
| Yes | 783 | 49.1 | 737 | 49.2 | 382 | 47.4 |
| Missing | 2 | 0.01 | — | — | — | — |
| Prepregnancy body mass index | ||||||
| Underweight (<18.5 kg/m 2 ) | 65 | 3.9 | 59 | 3.8 | 25 | 3.3 |
| Normal (18.5-24.9 kg/m 2 ) | 609 | 46.6 | 581 | 46.6 | 295 | 47.4 |
| Overweight (25.0-29.9 kg/m 2 ) | 303 | 23.0 | 290 | 23.2 | 163 | 24.3 |
| Obese (≥30 kg/m 2 ) | 394 | 26.4 | 372 | 26.4 | 185 | 25.1 |
| Smoking | ||||||
| No smoking during pregnancy | 979 | 72.5 | 933 | 72.5 | 494 | 75.6 |
| Stopped smoking before enrollment | 132 | 9.9 | 125 | 9.9 | 55 | 8.7 |
| <One-half pack/d | 182 | 11.5 | 167 | 11.4 | 84 | 9.6 |
| ≥One-half pack/d | 78 | 6.1 | 77 | 6.3 | 35 | 5.9 |
| Pregnancy | ||||||
| Gestational age at enrollment, wk | ||||||
| 15-19.9 | 225 | 15.8 | 211 | 15.6 | 102 | 14.1 |
| 20-24.9 | 964 | 71.1 | 915 | 71.2 | 482 | 73.2 |
| 25-27.9 | 182 | 13.2 | 176 | 13.3 | 84 | 12.6 |
| Fetal growth | ||||||
| Small for gestational age | 150 | 9.4 | 138 | 9.3 | — | — |
| Not small for gestational age | 1219 | 90.5 | 1162 | 90.6 | 668 | 100 |
| Missing | 2 | 0.1 | 2 | 0.1 | — | — |
| Hypertensive disorders | ||||||
| Gestational hypertension/preeclampsia a | 100 | 6.6 | 100 | 6.9 | — | — |
| Chronic hypertension | 49 | 3.0 | 45 | 3.0 | — | — |
| Normotensive | 1222 | 90.3 | 1157 | 90.0 | 668 | 100.0 |
a Preeclampsia includes all cases, even those superimposed on chronic hypertension.
Associations among maternal characteristics and angiogenic marker levels are presented without adjustment ( Tables 2 and 3 : total sample and normal pregnancies, respectively) and with adjustment ( Tables 4 and 5 : total sample and normal pregnancies, respectively). The text focuses on findings from the normal pregnancy sample.
| Characteristic | n | Placental growth factor | Soluble endoglin | Soluble Fms-like tyrosine kinase-1 | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Mean a | 95% CI | P value b | Mean a | 95% CI | P value b | Mean a | 95% CI | P value b | ||
| Gestational age at enrollment, wk | ||||||||||
| 15-19.9 | 211 | 228.2 | 207.9–250.4 | Reference | 5.30 | 5.06–5.55 | Reference | 1670 | 1504–1855 | Reference |
| 20-24.9 | 915 | 381.5 | 363.6–400.3 | < .01 | 5.28 | 5.17–5.40 | 1669 | 1593–1748 | ||
| 25-27.9 | 176 | 556.8 | 509.4–608.6 | < .01 | 5.81 | 5.51–6.13 | < .01 | 1676 | 1501–1871 | |
| Parity | ||||||||||
| Primiparous | 543 | 378.5 | 354.6–403.9 | Reference | 5.61 | 5.45–5.78 | Reference | 1948 | 1839–2063 | Reference |
| Multiparous | 759 | 364.7 | 344.9–385.7 | 5.18 | 5.06–5.30 | < .01 | 1501 | 1425–1581 | < .01 | |
| Maternal age at enrollment, y | ||||||||||
| <20 | 227 | 428.9 | 386.4–476.0 | Reference | 5.59 | 5.28–5.91 | Reference | 2002 | 1820–2201 | Reference |
| 20-29 | 734 | 358.1 | 339.3–378.0 | < .01 | 5.29 | 5.17–5.41 | 1586 | 1503–1673 | < .01 | |
| ≥30 | 341 | 367.0 | 336.4–400.5 | < .05 | 5.36 | 5.16–5.56 | 1688 | 1571–1814 | < .01 | |
| Medicaid | ||||||||||
| No | 565 | 354.0 | 333.1–376.2 | Reference | 5.46 | 5.32–5.61 | Reference | 1702 | 1609–1802 | Reference |
| Yes | 737 | 387.9 | 365.6–411.6 | < .05 | 5.24 | 5.11–5.38 | < .05 | 1637 | 1550–1730 | |
| Education, grade | ||||||||||
| <12 | 296 | 449.5 | 414.6–487.3 | < .01 | 5.32 | 5.11–5.54 | 1704 | 1550–1874 | ||
| 12 | 367 | 347.1 | 317.8–379.1 | Reference | 5.39 | 5.20–5.60 | Reference | 1631 | 1513–1757 | Reference |
| >12 | 639 | 357.5 | 337.5–378.8 | 5.34 | 5.21–5.48 | 1678 | 1590–1771 | |||
| Prepregnancy body mass index | ||||||||||
| Underweight | 59 | 555.4 | 470.8–655.2 | < .01 | 5.33 | 4.78–5.94 | 1939 | 1604–2345 | ||
| Normal | 581 | 412.6 | 387.3–439.7 | Reference | 5.57 | 5.42–5.72 | Reference | 1884 | 1782–1993 | Reference |
| Overweight | 290 | 378.0 | 347.3–411.3 | 5.43 | 5.25–5.62 | 1633 | 1513–1764 | < .01 | ||
| Obese | 372 | 283.4 | 263.8–304.4 | < .01 | 4.93 | 4.75–5.12 | < .01 | 1347 | 1245–1458 | < .01 |
| Smoking | ||||||||||
| No smoking during pregnancy | 933 | 350.2 | 333.7–367.5 | Reference | 5.41 | 5.30–5.53 | Reference | 1700 | 1623–1781 | Reference |
| Stopped smoking before enrollment | 125 | 317.6 | 275.2–366.5 | 5.47 | 5.14–5.83 | 1618 | 1420–1844 | |||
| <One-half pack/d | 167 | 507.3 | 451.8–569.7 | < .01 | 5.17 | 4.98–5.38 | < .05 | 1628 | 1451–1826 | |
| ≥One-half pack/d | 77 | 509.4 | 437.6–593.0 | < .01 | 4.83 | 4.40–5.30 | < .05 | 1493 | 1277–1745 | |
| Race | ||||||||||
| White/other | 766 | 348.9 | 330.9–367.9 | Reference | 5.31 | 5.18–5.43 | Reference | 1591 | 1514–1673 | Reference |
| African American | 536 | 444.2 | 419.6–470.2 | < .01 | 5.50 | 5.36–5.64 | < .05 | 1936 | 1839–2038 | < .01 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
