Preeclampsia is one of the major pregnancy complications, especially in nulliparas. Its etiology remains unknown. Knight et al.1 found that syncytiotrophoblast microvilli were shed into the maternal circulation and were present in significantly higher amounts in preeclamptic women. This research group has also demonstrated that the syncytiotrophoblast microvillous membranes from normotensive women inhibit the proliferation of endothelial cells, disrupt their growth as a monolayer in culture,2 and inhibit the endothelial cell-dependent relaxation of small arteries in vitro.3 These researchers hypothesized that the syncytiotrophoblast microvilli may contribute to the endothelial dysfunction underlying the maternal syndrome of preeclampsia.1 More recently, other investigators confirmed that the concentration of cell-free fetal DNA in maternal circulation was several-fold higher in women with preeclampsia than that in normotensive women4,5 and that this difference was already apparent early in the second trimester.6
In parallel studies, it was found that women who carried a fetus with trisomy 21 had a more than 2-fold increase in cell-free fetal DNA7,8 and in nucleated fetal cells9 in their bloodstream compared with women who carried an euploid fetus. Furthermore, increased levels of human chorionic gonadotropin beta subunit,10 inhibin A, and activin A are found in the serum of mothers who carry a trisomic fetus11 as well as of mothers who will eventually develop preeclampsia.12,13
If fetal antigen in the maternal circulation predisposes pregnant women to preeclampsia, then mothers who have a fetus with trisomy 21 should be more susceptible to developing preeclampsia. Conversely, if these women do not show a higher rate of preeclampsia, then either fetal antigen is not a cause or factors present in trisomic pregnancies are protective against disorder. To test this hypothesis, we examined the association between fetal trisomy 21 and maternal hypertensive disorders in pregnancy.
We used 2 data sources to examine this association. The first is the U.S. Natality data, which include a large number of Down syndrome births. However, the Natality data do not separate preeclampsia from gestational hypertension. Because the etiology and pathogenesis of preeclampsia and gestational hypertension may be distinct, we used another population-based case-control study to separate preeclampsia from gestational hypertension.
The U.S. Natality Data
U.S. Natality Data, compiled by the National Center for Health Statistics, record all livebirths in the United States. The birth certificate includes demographic and residential information, pregnancy complications, labor and delivery procedures, and information on the newborn. Detailed description of the birth certificate has been provided elsewhere.14 We included all livebirths in the United States from 1995–1999 (approximately 20 million births). We first excluded subjects with the following conditions: (1) multiple gestation; (2) unknown conditions for Down syndrome, hypertension in pregnancy, or chronic hypertension; (3) maternal chronic hypertension; (4) gestational age at birth less than 25 weeks; and (5) unknown parity. After these exclusions, all pregnancies with Down syndrome were selected as the exposed subjects (N = 7756) and all others as potential controls.
We then matched each trisomic pregnancy with 2 random euploid pregnancies based on the following variables: year and state/county of birth, hospital birth (yes/no), parity (nullipara/multipara), time prenatal care started (1st, 2nd, or 3rd trimesters; none; unknown), and maternal age (in years). Any exposed subject who failed to have 2 controls in the first match (91 exposed with no controls and 70 exposed with only one control) was rematched to the unmatched control population by modifying the prenatal care categories (combining none and unknown) and by using maternal age in 3 year groups, keeping the other variables unchanged. After the second match, 83 trisomic pregnancies remained unmatched and 50 had only one control pregnancy. The 83 unmatched subjects were then deleted. In the end, we have 7673 trisomic pregnancies and 15,293 matched euploid pregnancies.
In addition to the matched variables, the following variables were examined: fetal sex, marital status (married vs. others), maternal education (years in school), maternal race/ethnicity (non-Hispanic white, non-Hispanic black, Hispanics, and other), number of spontaneous or induced abortions, smoking during pregnancy (none, 1–9, 10–19, and 20+ cigarettes per day), gestational age at delivery (based on best clinical estimate), and birthweight. Two items related to hypertension in pregnancy were available in the database: pregnancy-associated hypertension and eclampsia. Given the rarity of eclampsia, we combined the latter 2 into “pregnancy-induced hypertension.” We examined the association between fetal trisomy 21 and maternal pregnancy-induced hypertension in this matched cohort.
The Population-based Case-control Study
From July 1991 through December 1993, a population-based case-control study on Down syndrome was conducted by the California Birth Defects Monitoring Program, an active surveillance program in selected counties that captures more than half of the births in California.15 Abstractors inspect the obstetric, nursery, pediatric, and pathology logs of all birthing and pediatric tertiary-care hospitals in the monitored counties and abstract the charts of all liveborn or stillborn infants who have a structural birth defect. Ascertainment of infants with birth defects continues until they reach 1 year of age. Abstractors also visit all genetic laboratories serving the monitored counties to record all abnormal karyotypes. Spontaneous and elective abortions with abnormal karyotypes are ascertained from this source or from hospital logs.
Infants and fetuses with ascertained Down syndrome, born in the surveillance area, were included as cases. Mothers of the cases were approached for participation in the study. Infants who were adopted or had left California, or whose mother did not speak English or Spanish, were excluded. Controls were the mothers of liveborn infants without a birth defect who were randomly selected from the surveillance area during the study period. There were 743,565 livebirths in the catchment region during that period. A total of 1197 cases and 1296 controls were eligible for the study. Ten percent of the cases and 16% of the controls could not be located. An additional 6% of cases and controls declined to participate. After excluding 7 mothers with incomplete interviews, a total of 997 (83%) cases and 1007 (78%) controls were included in the study. Because preeclampsia usually occurs in the second half of pregnancy, we excluded all subjects with a duration of gestation less than 26 weeks, as well as multiple gestations, leaving 674 trisomic cases and 993 euploid controls for analysis. Among those who were excluded, 83% were therapeutic abortions.
Mothers of cases and controls were interviewed in English or Spanish by telephone using a structured questionnaire. It includes maternal characteristics, reproductive and medical history, and the index pregnancy history. Information on hypertension in pregnancy was collected as follows: women were asked if they had ever been diagnosed as having hypertension or high blood pressure; in what month of pregnancy hypertension first occurred or whether it started before pregnancy and continued during the pregnancy; whether they took any prescribed medication; and what medicine they took before and during pregnancy. The same questions were asked for preeclampsia and eclampsia. We defined gestational hypertension as no hypertension before pregnancy or in the first trimester, but hypertension starting in the second or the third trimester with no preeclampsia. Preeclampsia was defined as hypertension plus proteinuria starting in the second or the third trimester and no hypertension before pregnancy or in the first trimester. Nine trisomic and 6 euploid pregnancies had missing information on hypertension and therefore were excluded.
This study was originally designed as a case-control study. However, the population-based nature of the study and the clear temporal sequence of events make it appropriate to treat this as a frequency-matched cohort study for the current analysis. Down syndrome is considered as the exposure and hypertensive disorders in pregnancy as the outcome. This study was approved by the Institutional Review Board of the California Birth Defects Monitoring Program.
For both the Natality data and the population-based study, we first examined the characteristics of subjects stratified by parity and exposure (euploidy vs. trisomy). We then examined the association between exposure (trisomy 21) and outcome (hypertension in pregnancy) adjusting for potential confounders. For the matched Natality data, we used generalized estimating equations (GEE) to estimate the adjusted relative risk (aRR) and 95% confidence interval (CI). We used regular logistic regression for the California study. We used a backward selection process to eliminate variables that had P values greater than 0.2 and changed the main effect (β coefficient) by less than 10%.
The U.S. Natality Data
Table 1 shows that there were slightly more non-Hispanic whites and Hispanics among trisomic pregnancies than among euploid pregnancies. Babies with Down syndrome had a shorter gestational age (by 1 week on average) and lower birthweight (by 340 g). These differences were consistent for nulliparous and multiparous mothers. Other maternal characteristics were very similar between the trisomic and euploid pregnancies. However, among nulliparous mothers, the incidence of pregnancy-induced hypertension was 4.4% in trisomic pregnancies and 6.5% in euploid pregnancies. The corresponding incidences of pregnancy-induced hypertension in multiparous women were 3.4% and 3.1%, respectively.
Because the distribution of gestational age differs between trisomic and euploid pregnancies, we used direct standardization method with the distribution in euploid pregnancies as the standard to calculate the standardized incidence of pregnancy-induced hypertension in trisomy and euploid pregnancies. For nulliparas, the standardized incidences were 3.6% and 6.5% for trisomic and euploid pregnancies, respectively, whereas for multiparas, the incidences were identical at 3.1%. GEE analysis, controlling for maternal race with or without stratification by gestational age, confirmed these patterns (for nulliparas, aRR = 0.67; 95% CI = 0.53 to 0.85; for multiparas, aRR = 1.12; CI = 0.93 to 1.35).
The Population-based Case-control Study
Table 2 presents the characteristics of pregnancies with trisomic and euploid fetuses, grouped by parity. Among nulliparous women, approximately half of the study population were Hispanic; most women were married and spoke English; more than one third had some college education; and the majority of women started prenatal care in the first trimester. A higher percentage of women with a trisomic fetus were Hispanic, had been interviewed in Spanish, and were married compared with women who had a euploid fetus. No major differences in maternal age, body mass index, smoking during pregnancy, diabetes, or thyroid disease were found between trisomic and euploid nulliparous pregnancies. Among multipara, women having trisomic pregnancies were more likely to be Hispanic and less educated; to have been interviewed in Spanish; to have unplanned pregnancies; and to have started prenatal care late. They were also older and had a higher body mass index. In both nullipara and multipara, pregnancies with trisomy 21 were delivered earlier than those with euploidy.
Among nulliparous women carrying a trisomy 21 fetus, the incidence of gestational hypertension and preeclampsia was lower. After adjusting for gravidity (primigravida or multigravida), maternal age (<23, 23 to 28, >28 years), race/ethnicity (Hispanic or non-Hispanic), marital status (yes or no), fertility treatment (yes or no), gestational age at delivery (<37, 37 to 38, >38 weeks), body mass index (continuous), and diabetes (yes or no), we found that fetal trisomy 21 was associated with a substantially reduced risk of maternal preeclampsia (aRR = 0.19; 95% CI = 0.04 to 0.88) (Table 3). However, the risk of preeclampsia was only modestly reduced in multiparous women with a trisomic fetus. Similarly, there were only minor and inconsistent associations between fetal trisomy 21 and gestational hypertension in nulliparas and multiparas. These results remained virtually the same when we restricted our analysis to those who were interviewed in English (results not shown).
We used 2 population-based datasets to examine the association between fetal trisomy 21 and maternal hypertension in pregnancy. Results of the first large dataset showed that in nulliparous women, fetal trisomy 21 was associated with a reduced risk of maternal pregnancy-induced hypertension. Findings from the second dataset confirmed this association and further revealed that the decrease in overall risk of pregnancy-induced hypertension was mainly the result of a sizable reduction in the risk of preeclampsia. In contrast, both datasets showed no consistent or important decreases in risk for pregnancy-induced hypertension among multiparous women.
The literature has suggested substantial differences in the etiology and pathogenesis of gestational hypertension and preeclampsia.16 Furthermore, hypertensive disorders in multiparous gestation may be more likely to be the result of underlying chronic conditions.16 Thus, our findings in distinguishing between gestational hypertension and preeclampsia and between nulliparas and multiparas are consistent with the current literature.
A trisomic gestation has a higher rate of early fetal loss and premature birth than a euploid pregnancy,17 ie, the entire distribution of gestational age shifts to the left. As the clinical manifestation of pregnancy-induced hypertension becomes more and more common with advancing gestation, it could be that the reduced risk of preeclampsia in nulliparous trisomic pregnancies was the result of the shorter gestational age. Our data did not support this concern. Both datasets showed that trisomic pregnancies had a shorter gestational age than euploid pregnancies by approximately 1 week. However, the reduced risk was observed only in nulliparas, not in multiparas. Furthermore, studies have shown that not only is the incidence of pregnancy-induced hypertension much lower in multiparas, but the clinical manifestation appears milder and occurs later in gestation.16 If a shorter gestational age had biased the results, one would expect to see a similar or even stronger reduction in multiparas. The discrepancy in results between nulliparas and multiparas also argues against the notion of a survival effect; healthy trisomic pregnancies surviving to late gestation might also have a lower risk of pregnancy-induced hypertension or preeclampsia. Finally, some women with a trisomic gestation elected to terminate pregnancy early. Such a decision appears unlikely to have any substantial relationship with preeclampsia in late pregnancy.
Histopathologic analysis of trisomy 21-affected placenta shows trophoblastic hypoplasia and villous hypovascularity from the first to the third trimester.18–20 Furthermore, a defect in the differentiation of cytotrophoblast into syncytium has been demonstrated in trisomy 21-affected placenta.21 According to the current theory of pathogenesis of preeclampsia, a poorly developed placenta would likely increase the risk of preeclampsia. Why then did pregnancies with trisomy 21 have a reduced risk of preeclampsia?
Numerous studies have indicated that placental oxidative stress and increased placental lipid peroxide production play significant roles in the pathogenesis and pathophysiology of preeclampsia.22 We hypothesize that a reduced risk of preeclampsia in pregnancies with trisomy 21 may be related to a protective effect of superoxide dismutase (SOD). SOD catalyzes dismutation of superoxide anion into oxygen and hydrogen peroxide. It is one of the key antioxidant enzymes for cellular defense against oxidative damage by reactive oxygen species. Copper/zinc SOD (SOD1) is present in the cytosol of virtually all animal cells. The SOD1 gene is located on chromosome 21q22.1.23 Compared with placentas from a normal pregnancy, placentas from preeclamptic women were found to have a decreased SOD activity and mRNA expression in trophoblast cells.24 In contrast, SOD1 mRNA expression, protein levels, and activity are significantly higher in trophoblast cells isolated from trisomy 21-affected placentas than in those from normal placentas.25 Thus, an extra copy of chromosome 21 may result in a higher level of SOD1 and, consequently, a protective effect from oxidative damage to the trophoblast and other placenta tissue, which ultimately may protect the mother from widespread endothelial injury and dysfunction.26
We acknowledge that neither dataset used in this study is perfect to test this hypothesis. First, Down syndrome and pregnancy-induced hypertension were likely to have been underreported in the Natality data. However, underreporting cannot explain the substantial discrepancy between nulliparas and multiparas. In the California study, the classification of hypertension in pregnancy was based on self-report, which is subject to error. Retrospectively reviewing medical records scattered in a large geographic region would be logistically challenging. On the other hand, we found that the incidence of gestational hypertension and preeclampsia among the euploid pregnancies in the current study (8.0% and 4.9% for nullipara, and 5.5% and 1.7% for multipara, respectively) is consistent with values reported for the general population.27 Conventional wisdom suggests that women with a bad perinatal outcome are more inclined to report complications than women with a normal outcome. If there were a significant reporting bias, one might expect to see a consistently higher incidence of the various disorders listed in Table 2 in pregnancies with a trisomic fetus; such a pattern was not observed. Therefore, although reporting bias and misclassification of preeclampsia are possible, they are unlikely explanations for our finding. When we restricted our analysis to subjects interviewed in English, assuming that language might be another potential source of bias, they did not affect our results.
The Natality data included only livebirths. In the California study, euploid pregnancies were restricted to livebirths, whereas trisomic pregnancies included stillbirths (4 of 665, one of which was to a mother with gestational hypertension). Because severe preeclampsia is associated with stillbirth, excluding stillbirth could have underestimated the incidence of preeclampsia. However, because the incidence of stillbirth is rare (<1%), even among severe preeclampsia (<4%),28 excluding stillbirth is likely to have a negligible impact on our finding. Finally, the number of preeclampsia cases in the California study is still small in some categories.
In summary, the risk of preeclampsia appears to be reduced in nulliparous pregnancies with a trisomy 21 fetus. We hypothesize that heightened expression and activity of copper/zinc superoxide dismutase in trisomy 21-affected placenta may have protective effects on preeclampsia.
We thank Enrique Schisterman for valuable discussions on methodology, Ann Trumble for computer programming support, and Berthold Huppertz and James Troendle for their helpful comments on the manuscript.
1. Knight M, Redman CWG, Linton EA, et al. Shedding of syncytiotrophoblast microvilli into maternal circulation in preeclamptic pregnancies. Br J Obstet Gynecol
2. Smarason AK, Sargent IL, Starkey PM, et al. The effect of placental syncytiotrophoblast microvillous membranes from normal and preeclamptic women on the growth of endothelial cells in vitro. Br J Obstet Gynecol
3. Cockell AP, Learmont JG, Smarason AK, et al. Human placental syncytiotrophoblast microvillous membranes impair maternal vascular endothelial cell function. Br J Obstet Gynecol
4. Lo YMD, Leung TN, Tein MSC, et al. Quantitative abnormalities of fetal DNA in maternal serum in preeclampsia. Clin Chem
5. Zhong XY, Laivuori H, Livingston JC, et al. Elevation of both maternal and fetal extracellular circulating deoxyribonucleic acid concentrations in the plasma of pregnant women with preeclampsia. Am J Obstet Gynecol
6. Leung TN, Zhang J, Lau TK, et al. Increased maternal plasma fetal DNA concentrations in women who eventually develop preeclampsia. Clin Chem
7. Lo YMD, Lau TK, Zhang J, et al. Increased fetal DNA concentrations in the plasma of pregnant women carrying fetuses with trisomy 21. Clin Chem
8. Zhong XY, Burk MR, Troeger C, et al. Fetal DNA in maternal plasma is elevated in pregnancies with aneuploid fetuses. Prenat Diagn
9. Bianchi DW, Williams JM, Sulllivan LM, et al. PCR quantitation of fetal cells in maternal blood in normal and aneuploid pregnancies. Am J Hum Genet
10. Brizot ML, Jauniaux E, Mckie AT, et al. Placental expression of alpha and beta subunits of human chorionic gonadotrophin in early pregnancies with Down's syndrome. Hum Reprod
11. Dalgliesh GL, Aithen DA, Lyall F, et al. Placental and maternal serum inhibin-A and activin-A levels in Down's syndrome pregnancies. Placenta
12. Sorensen TK, Williams MA, Zingheim RW, et al. Elevated second-trimester human chorionic gonadotropin and subsequent pregnancy-induced hypertension. Am J Obstet Gynecol
13. Muttukrishna S, North RA, Morris J, et al. Serum inhibin A and activin A are elevated prior to the onset of preeclampsia. Hum Reprod
14. Taffel SM, Ventura SJ, Gay GA. Revised US certificate of birth: new opportunities for research on birth outcome. Birth
15. Torfs CP, Christianson RE. Effect of maternal smoking and coffee consumption on the risk of having a recognized Down syndrome pregnancy. Am J Epidemiol
16. Lindheimer MD, Roberts JM, Cunningham FG. Chesley's Hypertensive Disorders in Pregnancy
, 2nd ed. Stamford, CT: Appleton & Lange; 1999.
17. Cuckle H. Down syndrome fetal loss rate in early pregnancy. Prenat Diagn
18. Qureshi F, Jacques SM, Johnson MP, et al. Trisomy 21 placentas: histopathological and immunohistochemical findings using proliferating cell nuclear antigen. Fetal Diagn Ther
19. Jauniaux E, Hustin J. Chromosomally abnormal early ongoing pregnancies: correlation of ultrasound and placental histological findings. Hum Pathol
20. Rochelson B, Kaplan C, Guzman E, et al. A quantitative analysis of placental vasculature in the third-trimester fetus with autosomal trisomy. Obstet Gynecol
21. Frendo JL, Vidaud M, Guibourdenche J, et al. Defect of villous cytotrophoblast differentiation into syncytiotrophoblast in Down syndrome. J Clin Endocrinol Metab
22. Roberts JM. Endothelial dysfunction in preeclampsia. Sem Reprod Endocrinol
23. Forsberg L, de Faire U, Morgenstern R. Oxidative stress, human genetic variation and disease. Arch Biochem Biophys
24. Wang Y, Walsh SW. Increased superoxide generation is associated with decreased superoxide dismutase activity and mRNA expression in placental trophoblast cells in preeclampsia. Placenta
25. Frendo JL, Therond P, Bird T, et al. Overexpression of copper zinc superoxide dismutase impairs human trophoblast cell fusion and differentiation. Endocrinology
26. Walsh SW. Maternal-placental interactions of oxidative stress and antioxidants in preeclampsia. Semin Reprod Endocrinol
27. Cunningham FG, Gant NF, Leveno KJ, et al. Williams Obstetrics
, 21st ed. New York: McGraw-Hill; 2001:572.
28. Zhang J, Meikle S, Trumble A. Preeclampsia-eclampsia associated severe maternal morbidity in the United States. Hypertens Pregnancy