Objective
The purpose of this study was to examine whether longitudinally sampled maternal angiogenic concentrations predict preeclampsia.
Study Design
Plasma sFlt-1 and placental growth factor (PlGF) concentrations in healthy pregnant women were quantified at 10, 17, 25, and 35 weeks’ gestation. Preeclampsia was diagnosed with criteria from the American College of Obstetricians and Gynecologists.
Results
In the first trimester, sensitivity/specificity for PlGF and sFlt-1 were 55/43% and 57/40%, respectively, and did not improve appreciably as the pregnancy progressed. Among pregnancies that later experienced preeclampsia, median PlGF was lower beginning in the second trimester, but sFlt-1 was not higher until the third trimester. Analyte positive predictive values approached 10% in the third trimester. Negative predictive values were >90% for the entire pregnancy.
Conclusion
Prediction of preeclampsia in early pregnancy was not possible with the use of maternal angiogenic protein concentrations. Even in late pregnancy, positive predictive values were not useful clinically. Negative predictive values are similarly unlikely to prove useful as a tool with which to a rule out suspected disease.
The pathogenesis of preeclampsia is hypothesized to be a multistep process. Most authorities agree that the first step involves a reduced cytotrophoblastic invasion of the uterine spiral arteries with a consequent decrease in the caliber of the arteries among patients with preeclampsia. Recent evidence suggests that, rather than hypoperfusion, the narrowed spiral arteries result in an increase in the velocity of blood that enters the intervillous space in the preeclamptic pregnancy. This higher velocity is associated with hypoperfusion-reperfusion injury to the fetal surface of the placenta and a resultant increase in oxidative stress. This stress is associated with the placenta release of soluble angiogenic proteins into the maternal systemic circulation. The clinical presentation of the disease is associated with marked abnormalities in the concentrations of these soluble factors Aberration in placentation is therefore key to the sequence of events that leads to the development of the disease.
Among the best characterized of the circulating placental angiogenic proteins are vascular endothelial growth factor (VEGF) and placental growth factor (PlGF), both of which interact with the family of VEGF receptors. The placenta is also the source of multiple circulating antiangiogenic proteins. Among these, the soluble isoform of VEGF receptor (sFlt-1) and soluble endoglin, which is a coreceptor for transforming growth factor beta, are best described at present. Elevated concentrations of sFlt-1 and soluble endoglin with simultaneously reduced concentrations of PlGF have been shown to precede the development of preeclampsia symptoms. Preeclampsia therefore appears to be a state of angiogenic imbalance with associated microangiopathic endothelial dysfunction within specific vascular beds, specifically those of the kidney, central nervous system, and liver. Several groups recently have suggested that the ratio of sFlt-1 to PlGF could be used as an alternative diagnostic criterion for preeclampsia.
The search for early prognostic biomarkers for preeclampsia has been a focus of recent research because this would allow the identification of patients whose condition would require additional clinical intervention and further elucidation of disease pathophysiologic events. Because the cytotrophoblastic invasion of the spiral arteries is believed to be completed by 18 weeks’ gestation, the initial pathophysiologic events that lead to preeclampsia are speculated to begin in the first or early second trimester when the initial angiogenic protein imbalance may have already commenced. If this model is correct, then increased concentrations of sFlt-1 and reduced concentrations of PlGF should exist in the first and early second trimester in women who will later experience the clinical manifestations of preeclampsia. We hypothesize that not only are aberrations in angiogenic proteins present in early pregnancy but also that these imbalances will allow the stratification of risk for the later development of clinical preeclampsia.
Materials and Methods
Study subjects
Participants were enrolled at 3 tertiary care academic centers: The Brigham & Women’s Hospital and the Beth Israel Deaconess Medical Center in Boston, MA, and The Hospital of the University of Pennsylvania in Philadelphia, PA. Eligible for study were women who had routine prenatal care at <15 weeks’ gestation, who were >18 years old, and who planned to deliver at the enrolling institution. The only exclusion criterion was higher-order multiple gestations (triplets or greater). Women with diabetes mellitus and chronic hypertension were excluded from the present analysis. The study protocol was approved by institutional review boards at each institution, and written informed consent was obtained from all participating women.
A total of 2246 singleton gestations with delivery at ≥24 weeks’ gestation were enrolled at the 3 study sites between October 2007 and June 2009. All subjects were enrolled prospectively in the first trimester, which is well before the development of any potential hypertensive complications of pregnancy. Among the 3 sites, the Brigham & Women’s Hospital contributed the most participants (48%); Beth Israel Deaconess Medical Center and the Hospital of the University of Pennsylvania contributed 29% and 23%, respectively. Study visits occurred at the following median (interquartile range) weeks of gestation: 9.7 (8.4–11.6), 17.8 (16.8–18.7), 25.9 (24.8–28.1), and 35.1 (34.6–35.9).
Biospecimen collection and processing
Maternal blood and urine samples were obtained at the 4 visits during the pregnancy. Approximately 10 mL of blood was drawn in ethylenediaminetetraacetic acid plasma tubes at each study visit; the samples were kept at +4°C until processing for storage within 4 hours of venipuncture. The specimens were centrifuged for 20 minutes and stored at –80°C. Samples were shipped in batches on dry ice to Abbott Diagnostics (Abbott Park, IL) where they were stored at –80°C until analysis.
Laboratory assays
PlGF and sFlt-1 were measured with prototype ARCHITECT immunoassays (Abbott Laboratories, Abbott Park, IL). The PlGF immunoassay measures the free form of PlGF-1. The assay has a lower limit of detection of 1 pg/mL, with a range up to 1500 pg/mL. The sFlt-1 immunoassay measures both free and bound sFlt-1. The assay has a lower limit of detection of 0.10 ng/mL with a range up to 150 ng/mL. The combined intra- and interassay coefficients of variation are <7% for PlGF and sFlt-1.
Clinical data
Information on the index pregnancy and neonate was abstracted from the medical record and supplemented with data that were collected specifically for the study. Maternal blood pressure and urine protein dip measurements were recorded at each study visit. The dates and times of the highest recorded blood pressures in the pregnancy also were noted. When applicable, the dates and times of the results of 24-hour urine protein collections were recorded. Height and weight were recorded at the first study visit. Gestational age was confirmed by ultrasound scanning at <15 weeks’ gestation. Date and time of delivery, birthweight, gender, Apgar score, mode of delivery, diagnoses of any pregnancy complications, medication use, and conception by assisted reproductive technologies were abstracted from records.
Questionnaire data
The participants completed a brief questionnaire that ascertained information about race/ethnicity, tobacco use before and during the index pregnancy, medical history, and history of preeclampsia in a previous pregnancy.
Definition of preeclampsia and gestational hypertension
Gestational hypertension was defined as blood pressure ≥140 mm Hg systolic or ≥90 mm Hg diastolic at study visits 2-4 with negative urine protein test results. Preeclampsia was defined as gestational hypertension with positive urine protein test result (>300 mg/24 hours or protein/creatinine >0.20). The specifics of patient treatment were the responsibility of the participating institution; however, the specifics included inpatient treatment of both mild and severe preeclampsia with delivery at 37 weeks and 34 weeks’ gestation, respectively. All cases of hypertensive disease were deidentified and reviewed by a panel of the principle investigators from each center. A final diagnosis was assigned only with the approval of this panel. Based on these criteria, 202 pregnancies were identified as preeclamptic, and 246 pregnancies were identified with gestational hypertension.
Statistical methods
The conservative Wilcoxon Rank Sum test was used to compare medians for continuous variables between outcome groups because a normal distribution did not pertain.
Receiver operating characteristic (ROC) was used to calculate the optimal cutoffs for the angiogenic analytes. ROC is a graphic plot of the sensitivity or true-positive rate vs the false-positive rate (1-specificity or 1-true-negative rate), because a binary variable as its discrimination threshold is varied. The ROC also can be represented equivalently by a plot of the fraction of true positives of the test positives (true-positive rate) vs the fraction of false positives out of the test negatives (false-positive rate). Optimal cutoffs were determined by minimization of the distance between the ideal sensitivity (100%) and 1-specificity (0%) point and the observed sensitivity and 1-specificity in the ROC analysis.
To assess whether the associations of angiogenic proteins with preeclampsia were explained by other risk factors for preeclampsia, logistic regression models estimated the relative risk of preeclampsia with a comparison of women with values above and below the optimal cutoffs for each of the analytes and the PlGF/sFlt-1 ratio. Models included maternal tobacco use, body mass index, age, race/ethnicity, parity, and study site as independent variables. Statistical significance was defined as a probability value of < .05.
Results
Description of study sample
Table 1 presents the selected characteristics of the study cohort by preeclampsia status. Age at study enrollment (31-32 years) did not differ appreciably by hypertensive status; however, nulliparity was more common among women who were diagnosed with gestational hypertension and preeclampsia. A higher proportion of mothers with preeclampsia self-identified as African American, and a significantly lower proportion of mothers self-identified as white or Asian.
| Demographic | Clinical characteristic | ||||
|---|---|---|---|---|---|
| Uncomplicated (n = 1951; 87.0%) | Gestational hypertension (n = 153; 6.8%) | Preeclampsia (n = 139; 6.2%) | T otal (n = 2243) | P value | |
| Maternal age, y a | 31.9 (16.0–50.3) | 30.8 (18.0–50.2) | 31.4 (17.4–58.6) | 31.8 (16.0–50.3) | .0532 b |
| Nulliparity, n (%) | 551 (28.2) | 56 (36.6) | 55 (39.6) | 662 (29.5) | .0025 c |
| Race/ethnicity, n (%) | |||||
| White | 1185 (60.7) | 84 (54.9) | 64 (46.0) | 1333 (59.4) | .0015 c |
| African American | 384 (19.7) | 40 (26.1) | 57 (41.1) | 481 (21.4) | < .0001 c |
| Asian | 125 (6.4) | 4 (2.6) | 4 (2.9) | 133 (5.9) | .0466 |
| Hispanic | 187 (9.6) | 18 (11.8) | 12 (8.6) | 217 (9.7) | .6202 |
| Other | 65 (3.3) | 7 (4.6) | 1 (0.7) | 73 (3.3) | .1556 |
| Body mass index at first visit, kg/m 2 a | 24.0 (16.4–56.8) | 27.3 (17.9–56.9) | 27.4 (18.3–67.8) | 24.4 (16.4–67.8) | < .0001 b |
| Previous preeclampsia, n (%) | 62 (3.2) | 8 (5.2) | 16 (11.5) | 86 (3.8) | < .0001 |
| Any smoking in pregnancy, n (%) | 499 (25.6) | 41 (26.8) | 27 (19.4) | 567 (25.3) | .2466 |
| In vitro fertilization, n (%) | 80 (4.1) | 5 (3.3) | 11 (7.9) | 96 (4.3) | .0814 |
a Data are given as median (range);
b Based on Kruskal-Wallis test;
Mothers with preeclampsia had a significantly higher body mass index at first prenatal visit and were more likely to have had preeclampsia in a previous pregnancy. Although not statistically significant, a lower proportion of patients with preeclampsia reported smoking at any point before or during the pregnancy. Women who conceived by assisted reproductive technologies were over-represented among the preeclampsia pregnancies by a factor of almost 2, although this difference did not reach statistical significance.
Pregnancy and neonatal outcomes
As determined by study protocol, the gestational age at delivery was significantly lower among the preeclampsia pregnancies (median, 37.8 weeks’ gestation; interquartile range (ICR), 36.6–38.9 weeks’ gestation) vs the non-preeclampsia pregnancies (median, 39.1 weeks’ gestation; ICR, 38.3–40.0 weeks’ gestation; P < .0001). The more clinically meaningful rates of delivery at <37 weeks’ gestation (30.1% vs 8.6%; P < .0001) and at <34 weeks’ gestation (9.4% vs 1.9 %; P < .0001) also were higher among the preeclampsia pregnancies. Birthweight was significantly lower among the preeclampsia pregnancies (median, 2.95 kg; ICR, 2.5–3.4) vs non-preeclampsia pregnancies (median, 3.4 kg; ICR, 3.0–3.6; P < .0001). A higher percentage of babies from preeclampsia pregnancies had a birthweight Z-score less than –2 (15.4% vs 2.6 %; P < .0001). The proportions of male infants among the groups did not differ (47.0% vs 50.9%; P = .16).
Sequential analyte values
Table 2 displays values for PlGF, sFlt-1, and PlGF/sFlt-1 ratio by uncomplicated pregnancies and by those pregnancies diagnosed with gestational hypertension and preeclampsia. A clearly interpretable pattern regarding the relation between the analytes and the eventual hypertensive diagnosis is not apparent at the first study visit. By the second study visit at approximately 17 weeks’ gestation, however, patterns in the concentrations of both PlGF and sFlt-1 begin to emerge that persist for the remainder of the gestation. Women with uncomplicated pregnancies had the highest concentration of PlGF, those who experienced gestational hypertension had an intermediate concentration, and those who experienced preeclampsia had the lowest median concentration.
| Median wk | Analyte | Overall P value a | Clinical characteristic | |||||
|---|---|---|---|---|---|---|---|---|
| Uncomplicated b | P value c | Gestational hypertension b | P value d | Preeclampsia b | P value e | |||
| 10 | PlGF | .0324 | 20.7 (14.7–32.2) | .0225 | 17.8 (13.9–29.3) | .0128 | 21.0 (15.9–34.4) | .2675 |
| sFlt-1 | < .0001 | 5.1 (3.4–7.4) | < .0001 | 4.1 (2.6–5.6) | .0094 | 4.7 (3.6–6.8) | .3520 | |
| PlGF/sFlt-1 | .0158 | 4.4 (3.0–7.0) | .0147 | 5.5 (3.3–7.8) | .8198 | 5.0 (3.1–8.5) | .0891 | |
| 17 | PlGF | .0064 | 137.8 (98.2–193.0) | .0467 | 128.1 (90.3–183.8) | .4709 | 113.2 (79.2–196.7) | .0093 |
| sFlt-1 | < .0001 | 6.3 (4.2–9.4) | < .0001 | 5.1 (3.5–6.8) | .0021 | 6.0 (4.5–9.3) | .9572 | |
| PlGF/sFlt-1 | .0189 | 22.2 (13.2–35.9) | .0209 | 24.9 (16.7–39.0) | .0082 | 20.2 (12.0–33.0) | .1425 | |
| 24 | PlGF | < .0001 | 455.0 (308.5–672.7) | .0122 | 394.7 (253.7–565.9) | .0025 | 333.1 (199.7–489.8) | < .0001 |
| sFlt-1 | < .0001 | 6.0 (3.9–9.2) | < .0001 | 4.7 (2.9–7.4) | < .0001 | 6.5 (4.2–13.2) | .0340 | |
| PlGF/sFlt-1 | < .0001 | 77.7 (43.3–128.3) | .1659 | 84.0 (48.9–150.1) | < .0001 | 55.5 (27.4–88.0) | < .0001 | |
| 35 | PlGF | < .0001 | 385.6 (175.2–750.8) | .0001 | 256.9 (134.0–552.5) | < .0001 | 134.9 (65.2–308.5) | < .0001 |
| sFlt-1 | < .0001 | 9.7 (6.7–14.6) | .2528 | 11.0 (6.7–16.6) | < .0001 | 18.4 (10.4–32.0) | < .0001 | |
| PlGF/sFlt-1 | < .0001 | 42.0 (14.9–98.6) | .0008 | 26.1 (8.7–68.7) | < .0001 | 7.6 (2.7–30.1) | < .0001 | |
a Kruskall-Wallace tests, comparison of all groups;
b Data are given as median (interquartile range);
c Mann-Whitney statistical tests, comparing gestational hypertension vs uncomplicated;
d Mann-Whitney statistical tests, comparing gestational hypertension vs preeclampsia;
e Mann-Whitney statistical tests, comparing preeclampsia vs uncomplicated.
The pattern for sFlt-1 was different from that for PlGF. From the second study visit onward through pregnancy, sFlt-1 concentrations were similar for uncomplicated and gestational hypertension pregnancies. However, sFlt-1 concentrations increased progressively and dramatically among those pregnancies that were diagnosed with preeclampsia. The median PlGF/sFlt-1 ratio followed a similar pattern to s-Flt1 with regard to hypertensive disease category and stage of pregnancy.
Testing characteristics
Table 3 lists the results of the ROC analysis and the associated test parameters for each of the analytes and the analyte ratio by study visit. For each, the area under the curve increased with advancing gestational age, particularly at and after the third study visit at 24 weeks’ gestation. This information was used to calculate an optimal cutoff to maximize both sensitivity and specificity and positive and negative predictive values. Specificity improved with advancing gestational age, less so sensitivity. The positive predictive value remained relatively low across all study time points, whereas the negative predictive value, although high to begin with, improved from approximately 93% to approximately 97% with advancing gestation. Combining gestational hypertension with uncomplicated pregnancy in the comparison group did not change the results materially (data not shown).
| Median wk | Analyte | Area under the curve (95% CI ) | Optimal cutoff | Sensitivity, % | Specificity, % | Positive predictive value, % | Negative predictive value, % |
|---|---|---|---|---|---|---|---|
| 10 | PlGF | 0.53 (0.48–0.58) | ≤22.77 | 55.6 | 43.4 | 5.9 | 93.8 |
| sFlt-1 | 0.52 (0.47–0.57) | ≥4.33 | 57.1 | 39.5 | 5.7 | 93.5 | |
| Ratio: PlGF/sFlt-1 | 0.54 (0.49–0.59) | ≤5.10 | 51.1 | 42.7 | 5.4 | 93.3 | |
| 17 | PlGF | 0.57 (0.51–0.63) | ≤121.53 | 52.9 | 60.0 | 7.5 | 95.4 |
| sFlt-1 | 0.51 (0.45–0.56) | ≥6.70 | 47.1 | 55.4 | 6.1 | 94.5 | |
| Ratio: PlGF/sFlt-1 | 0.54 (0.49–0.60) | ≤23.94 | 61.3 | 46.6 | 7.0 | 95.2 | |
| 24 | PlGF | 0.66 (0.61–0.71) | ≤379.41 | 62.2 | 61.6 | 9.5 | 96.2 |
| sFlt-1 | 0.56 (0.51–0.62) | ≥7.09 | 46.5 | 61.8 | 7.3 | 94.7 | |
| Ratio: PlGF/sFlt-1 | 0.64 (0.59–0.69) | ≤60.23 | 52.8 | 61.4 | 8.1 | 95.3 | |
| 35 | PlGF | 0.72 (0.67–0.77) | ≤242.67 | 67.8 | 64.8 | 10.4 | 97.1 |
| sFlt-1 | 0.72 (0.67–0.77) | ≥13.12 | 62.7 | 68.9 | 10.9 | 96.8 | |
| Ratio: PlGF/sFlt-1 | 0.74 (0.69–0.79) | ≤14.42 | 61.0 | 75.1 | 13.0 | 97.0 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
