- Top of page
Objectives To determine the importance of genetic effects in the aetiology of pre-eclampsia and gestational hypertension and to investigate whether pre-eclampsia and gestational hypertension share genetic aetiology.
Design Individual record linkage between the population-based Swedish Multi-Generation and the Medical Birth Registers.
Population 1,188,207 births between 1987 and 1997 and their parents.
Methods Similarities in relatives were measured by the number of pairs concordant and discordant for disease, the odds ratio (OR) and tetrachoric correlations. Estimates of genetic and environmental effect for gestational hypertension, pre-eclampsia and pregnancy-induced hypertension were calculated from structural equation model fitting.
Main outcome measures Pre-eclampsia and gestational hypertension.
Results Full sisters and mother–daughters were more similar for pre-eclampsia (OR 3.3, 95% confidence interval [CI] 3.0–3.6 and OR 2.6, 95% CI 1.6–4.3, respectively) than half-sisters (maternal half-sisters OR 1.4, 95% CI 0.9–2.2 and paternal half-sisters OR 1.0, 95% CI 0.6–1.6). Full sisters and mother–daughters were also more similar for gestational hypertension than half-sisters. A full sister to a woman with pre-eclampsia also had a significantly increased risk of gestational hypertension (OR 2.5, 95% CI 2.2–2.8). In contrast, the risk for half-sisters was not increased. Model fitting suggested heritability estimates for pre-eclampsia of 31%, for gestational hypertension 20% and for pregnancy-induced hypertension 28%.
Conclusions There is a genetic component in the development of pre-eclampsia and gestational hypertension and the pattern of co-morbidity suggests that they may share part of their genetic aetiology. This could be important for studies of potential susceptibility genes for these diseases.
- Top of page
Pre-eclampsia and gestational hypertension develop in 2–10% of all pregnancies.1 Pre-eclampsia is associated with intrauterine growth restriction, preterm birth and maternal and perinatal deaths.2 Although gestational hypertension is often considered a less severe condition,3,4 the risks of adverse perinatal outcomes are higher in severe gestational hypertension than in ‘mild pre-eclampsia’.5 The primary aetiology of both conditions remains essentially unknown.6
There is evidence suggesting that not only pre-eclampsia but also gestational hypertension are partly genetic in origin. Maternal pre-eclampsia was associated with 70% excess risk of pre-eclampsia in daughters in a geographical area of Sweden.7 Results from other studies also support the hypothesis of a familial susceptibility to eclampsia and pre-eclampsia.8–16 Moreover, a maternal genetic component of the liability of developing pre-eclampsia has been identified.12,14,17 Several models of inheritance have been proposed, and specific candidate genes that may account for maternal susceptibility have been suggested.13,18,19 However, many genes appear to be involved, and there is no simple mode of inheritance. There is also a genetic susceptibility to gestational hypertension, although probably weaker than that for pre-eclampsia.20
Although there is accumulated evidence for genetic influences on pre-eclampsia, the magnitude of such effects remains largely unknown. The reason is that genetic influences are ordinarily estimated from twin studies, comparing co-morbidity in monozygotic twins (genetically identical) and dizygotic twins (who share 50% of their segregating genes). Due to the rarity of the disease, there is, to our knowledge, only one twin study that estimated the heritability for liability to pre-eclampsia (54%) and gestational hypertension (24%),20 whereas two other studies, using hospital records, could not find any concordant pairs.21,22 However, these studies were of limited size, and the precision of the genetic influence should be viewed with caution.
Another interesting issue is whether pre-eclampsia and gestational hypertension are different expressions of the same genetic or environmental liability. There are similarities in risk factor patterns of pre-eclampsia and gestational hypertension indicating similarities in the biological mechanisms of the development of these conditions.1 Moreover, women who developed pre-eclampsia in their first pregnancy have an increased risk of gestational hypertension in their second pregnancy.23
The unique nationwide Swedish Multi-Generation Register includes information on all first-degree relatives born in Sweden from 1941 and onward. We have used this and the Medical Birth Register to model the similarity between mother–daughter pairs and sister pairs (full sisters and half-sisters on maternal or paternal side). We estimated the relative importance of genetic and environmental effects on the liability of developing pre-eclampsia and gestational hypertension and also calculated the risk of co-morbidity.
- Top of page
The data were derived through record linkage between the Multi-Generation Register (held by Statistics, Sweden) and the Medical Birth Register (held by the Swedish National Board of Health and Welfare). Use of the individual national registration numbers, uniquely assigned to each Swedish resident, permitted linkage of information about individuals across the population-based registers.24
The Multi-Generation Register was created by Statistics Sweden in the early 1990s by linkage of several different data sources, providing information on all first-degree relatives for residents born in Sweden 1941 or later. To be included in the register, individuals had to be alive in 1960 or born thereafter. Adoptions and other non-biological relations are flagged. Individuals who immigrated after 1960 but died or emigrated before 1992 are not included. The same is true for individuals who immigrated after 1992.25
The Medical Birth Register covers 99% of all births in Sweden from 1973 and onwards and includes prospectively collected information about complications during pregnancy and delivery.26,27 In Sweden, prenatal care is standardised and free of charge. Registration to prenatal care generally occurs at 8–12 gestational weeks. Prenatal routines includes visits every fourth week up to 24 gestational weeks, then every second week to 36 weeks, and weekly thereafter. At each visit, blood pressure is measured and urine is checked for protein using a dipstick. More than 95% of the pregnant population attend antenatal care before the 15th gestational week, and 90% of the pregnant population make at least nine visits to antenatal care.28
Diagnoses during pregnancy and delivery are noted by the obstetrician at the time of discharge from hospital, using a guide sheet, where definitions of diagnoses are written in clear text beside the International Classification of Diseases (ICD) code and a checkbox. Diagnoses are classified according to the Swedish version of the ICD, ninth revision (ICD-9) from 1987 through 1996 and according to the tenth revision (ICD-10) from 1997 and onwards. Gestational hypertension was defined as blood pressure increase during pregnancy to at least 140/90 mmHg measured on at least two occasions occurring after 20 weeks of gestation (ICD-9 codes 642D and 642X; ICD-10 codes O13 and O16). Pre-eclampsia was defined as gestational hypertension combined with proteinuria (two urinary protein dipsticks of at least 1+ or 300 mg of protein or more in a 24-hour urine collection, ICD-9 codes 642E and 642F; and ICD-10 code O14). Pregnancy-induced hypertension was defined as either gestational hypertension or pre-eclampsia and analysed as an ordinal scale with three values: (1) unaffected individuals, (2) individuals with gestational hypertension and (3) individuals with pre-eclampsia. The accuracy of the diagnoses in the register for pre-eclampsia and gestational hypertension using ICD-9 codes has previously been evaluated. Among 148 pregnancies occurring in 1987–1993, 137 women with ICD-9 codes 642E or 642F had pre-eclampsia (positive predictive value = 93%), and among 115 pregnancies coded as gestational hypertension (ICD-9 codes 642D and 642X), 97 women had gestational hypertension according to the notes in the individual records (positive predictive value = 84%).1
All pregnancies during the study period (1987–1997) were used in order to define whether the women had had pre-eclampsia or gestational hypertension. If a woman had at least one pregnancy with pre-eclampsia, she was classified into that category. The same approach was applied to classify gestational hypertension. Sisters and mothers to women with pre-eclampsia or gestational hypertension were identified, and pairs of healthy, disease discordant and disease concordant women were made. A sibship that contains one affected and one unaffected woman counts as one discordant pair; a sibship with two affected women resulted in one concordant pair; a sibship with three affected women (sister A, sister B and sister C) was counted as three concordant pairs (AB, AC and BC); a sibship with two affected and one unaffected women was counted as one concordant and two discordant pairs, and so on. The sample consisted of 312,310 full sister pairs, 26,748 maternal half-sister pairs, 32,757 paternal half-sister pairs and 51,684 mother–daughter pairs.
The risk of pre-eclampsia for women whose sister had been diagnosed with pre-eclampsia, compared with women whose sister had not been diagnosed with pre-eclampsia, was calculated for the different types of sister pairs, and the same procedure was applied to gestational hypertension. The risk was estimated as an odds ratio (OR). For mother–daughter pairs, we estimated the risk of pre-eclampsia among daughters to mothers who had been diagnosed with pre-eclampsia, compared with the risk of daughters whose mother had not been diagnosed with pre-eclampsia (again, the same procedure was applied to gestational hypertension). Confidence intervals (95% CI) were estimated according to Mantel–Haenszel's method.29
In the model fitting analysis, pre-eclampsia and gestational hypertension were entered as binary traits, whereas pregnancy-induced hypertension was entered as an ordinal trait, assuming an underlying normal distribution of liability, with multiple factors contributing additively. This liability was the sum of genetic and environmental effects.30 The similarity between relatives, with respect to diagnosis was calculated as a tetrachoric or polychoric correlation (sometimes referred to as correlation in liability)31 and 95% CI were computed for each relation. The importance of genetic effects was indicated by higher relative risks and correlation among first-degree compared with second-degree relatives.
Quantitative genetic analyses can differentiate the familial aggregation into effects due to shared genes from those due to shared environments. Moreover, the method also estimates the magnitude of the different effects. A full description of the method is provided in Czene et al. 2002.32 In short, the number of healthy, disease concordant and disease discordant relative pairs were entered into the structural equation model fitting program (Mx, Virginia Commonwealth University, Richmond, Virginia).33 The analyses are based on the assumptions that (1) the correlation between full sisters and mother–daughter pairs depends on common genes (average 50%) and a common familial environment, (2) the correlation between half-sisters with the same mother depends on common genes (average 25%) and a common familial environment and (3) the correlation between half-sisters with the same father depends on common genes (average 25%) only, since most children continue to live with their mother.
To examine the nature of the familial aggregation we used the equation P=A+C+E, where P is the liability to disease, A is the genotype (heritability), C is the shared familial environment (assuming it is equal in mother–daughter and sister relations) and E is the non-shared environment. Shared familial environment can in turn be divided into sister-specific (Cs) and a common family component (Cc). The equation then becomes; P=A+Cc+Cs+E. The equations for the expected correlations are as follows:
Correlation for full sisters = 0.5A+Cc+Cs
Correlation for biological mother–daughter pairs = 0.5A+Cc
Correlation for maternal biological half-sisters = 0.25A+Cc+Cs
Correlation for paternal biological half-sisters = 0.25A
- Top of page
Of the total number of 1,188,207 births, 32,824 (2.8%) were found to have the diagnosis of pre-eclampsia. Compared with a woman whose full sister did not have pre-eclampsia, a woman whose full sister had pre-eclampsia had a more than threefold increased risk (Table 1). Similarly, the relative risk for a daughter to be diagnosed with pre-eclampsia given that the mother had the diagnosis was also significantly increased (OR 2.6, 95% CI 1.6–4.3). Among maternal half-sisters, where one sister in the pair was affected, the risk of developing pre-eclampsia for the healthy sibling was not statistically significant. Corresponding risk among half-sisters on father's side was close to unity. The genetic effects of pre-eclampsia were further emphasised by the tetrachoric correlations, which decreased with decreasing genetic similarity. The correlation for pre-eclampsia among full sisters was slightly stronger than that for mother–daughters.
Table 1. Number of concordant and discordant pairs, OR and tetrachoric correlation for pre-eclampsia and gestational hypertension among sister and mother–daughter pairs.
| ||Number of pairs||OR||95% CI||Tetrachoric correlation||95% CI|
|Concordant affected||Discordant||Concordant unaffected|
|Full sisters||418||12,368||299,524||3.3||3.0 to 3.6||0.22||0.20 to 0.25|
|Half-sisters (mother)||23||1290||25,435||1.4||0.9 to 2.2||0.06||−0.02 to 0.14|
|Half-sisters (father)||18||1507||31,232||1.0||0.6 to 1.6||0.00||−0.08 to 0.08|
|Mother–daughters||17||134a||49,107||2.6||1.6 to 4.3||0.15||0.06 to 0.24|
|Full sisters||99||5649||306,562||3.8||3.1 to 4.7||0.21||0.17 to 0.24|
|Half-sisters (mother)||6||488||26,254||2.7||1.2 to 6.0||0.15||0.00 to 0.28|
|Half-sisters (father)||4||627||32,126||1.3||0.5 to 3.5||0.04||−0.11 to 0.17|
|Mother–daughters||4||108a||50,985||3.2||1.2 to 8.8||0.14||0.11 to 0.16|
We also conducted analysis only including monozygotic twins (who share 100% of their segregating genes) in 928 monozygotic twin pairs. Compared with a monozygotic twin whose co-twin did not have pre-eclampsia, a monozygotic twin whose co-twin had pre-eclampsia faced a substantially increased risk of pre-eclampsia (OR 33.6, 95% CI 7.8–145.0).
The diagnosis of gestational hypertension was registered in 12,835 (1.1%) births. Compared with women whose full sister did not have gestational hypertension, we observed an almost fourfold increase in risk among women whose full sister had gestational hypertension (Table 1). For half-sisters with the same mother and mother–daughter pairs, corresponding risks were also increased, while the risk for half-sisters with the same father was considerably lower and did not reach statistical significance.
To study if there was a common aetiology of gestational hypertension and pre-eclampsia, we analysed the risk of developing pre-eclampsia for a woman whose relative was affected with gestational hypertension, and the risk of developing gestational hypertension for a woman whose relative was affected with pre-eclampsia (Table 2). Among women for whom the full sister had pre-eclampsia, there was a more than doubled increase in risk to develop gestational hypertension and vice versa(Table 2). We observed a similar risk increase for a daughter to be diagnosed with pre-eclampsia given that the mother had gestational hypertension, while no increased risk of gestational hypertension could be seen for daughters to mothers with pre-eclampsia (OR 0.6, 95% CI 0.1–4.1). No increase in risk of co-morbidity was obtained for maternal or paternal half-sisters (Table 2).
Table 2. OR and 95% CI for the co-morbidity of pre-eclampsia and gestational hypertension.
| ||OR||95% CI|
In Table 3, the quantitative genetic analyses for pre-eclampsia, gestational hypertension and pregnancy-induced hypertension overall are summarised. For pre-eclampsia, the genetic effects accounted for 31% (95%, CI 9–45) of the variation in liability to pre-eclampsia and the non-shared environmental effects accounted for 63% (95%, CI 55–74). For gestational hypertension, the genetic effects accounted for 20%, although not significant, and non-shared environment effects accounted for 69% (95%, CI 54–83) of the variation in liability to gestational hypertension. The estimates for shared environment were low and non-significant for both pre-eclampsia and gestational hypertension.
Table 3. Estimates of genetic and environmental effects for pre-eclampsia, gestational hypertension and pregnancy-induced hypertension from structural equation model fitting.
|Measured||Parametre estimates (95% CI)a||Fit of modelb|
|Pre-eclampsia||0.31 (0.09–0.45)||0.00 (0.00–0.00)||0.06 (0.00–0.13)||0.63 (0.55–0.74)||0.005|
|Gestational hypertension||0.20 (0.00–0.46)||0.06 (0.00–0.23)||0.05 (0.00–0.19)||0.69 (0.54–0.83)||0.000|
|Pregnancy-induced hypertension||0.28 (0.17–0.34)||0.00 (0.00–0.00)||0.07 (0.07–0.07)||0.64 (0.60–0.67)||0.005|
As the co-morbidity analyses indicated a common aetiology for pre-eclampsia and gestational hypertension, we also performed quantitative genetic analyses on pregnancy-induced hypertensive diseases, with pre-eclampsia and gestational hypertension as different degrees of disease severity. In these analyses the estimates of heritability, shared (sister specific) and non-shared environment were 28% (95%, CI 17–34), 7% (95%, CI 7–7) and 64% (95%, CI 60–67), respectively. Since only pregnancies between 1987 and 1997 were included, it was not possible to perform the analyses with sufficient power on only first-born children, which would have been desirable. Nevertheless, all analyses have been done on primiparas and the results remained essentially unchanged (data not shown).
- Top of page
We have shown a genetic component in the development of pre-eclampsia, which corroborates findings from previous smaller studies.14,20 We have estimated the importance of maternal genes to the liability to develop pre-eclampsia to be around 30%. We have also been able to show that there is a familial aggregation for gestational hypertension, which is probably also mainly due to genetic effects. As there also seems to be a genetic effect on co-morbidity between pre-eclampsia and gestational hypertension, our results further indicate that pre-eclampsia and gestational hypertension share genetic aetiology.
The genetic effect of pre-eclampsia was further emphasised by the measures of similarity (i.e. correlations and concordances), which decreased with decreasing genetic similarity, and the high risk for co-twins of affected monozygotic twins. The twin correlations were in agreement with the study by Salonen Ros et al.20 where monozygotic twins had a correlation of 0.6 and the dizygotic twins a correlation of 0.2.
The genetic effect accounted for 31% of the variation in liability to pre-eclampsia. This can be compared with the twin study of Salonen Ros et al.,20 where the maternal genetic effect accounted for 54% of the variation, and with 22% from a British study of self-reported pre-eclampsia,21 whereas two twin studies using medical records found no concordant twin pairs.21,22 It is clear that pre-eclampsia is a heritable trait, where genetic effects account for a sizeable proportion of the liability.
Studies on the genetic component to the liability of gestational hypertension are conflicting. Sutherland et al.14 did not find any familial tendency. Compared with women whose full sisters did not have gestational hypertension, a woman whose full sister had gestational hypertension had an almost fourfold increased risk. However, when quantifying the genetic effects, we found, in agreement with the results of Salonen Ros et al.,20 a non-significant genetic component of 20% of the variation in liability to gestational hypertension.
We considered whether pre-eclampsia and gestational hypertension are different expressions of the same genetic or environmental liability. We found that women whose sister developed pre-eclampsia had significantly increased risk (OR 2.5) of developing gestational hypertension and vice versa, suggesting a common familial tendency. These results prompted us to analyse pre-eclampsia and gestational hypertension as a trait with different degrees of disease severity. When quantitative genetic analyses were performed, the estimated genetic effect on the liability of pregnancy-induced hypertension was 28%, whereas corresponding genetic estimate in the twin study amounted to 47%.20 When a genome-wide scan was performed in Iceland including women with non-proteinuric and proteinuric pregnancy hypertension, a significant locus on 2p13 was detected, indicating a susceptibility locus in common.18 Previous reports suggest that a substantial fraction (46%) of women who develop gestational hypertension in early pregnancy also develop pre-eclampsia later during pregnancy, and around 10% progress to severe pre-eclampsia.34 This indicates that, to some extent, the diagnosis of a woman is dependent on how soon after the occurrence of hypertension the delivery takes place. Several differences have been described between gestational hypertension and pre-eclampsia, for example, level of pulse pressure in early pregnancy35 and haemodynamics.36c On the other hand, compared with ‘mild pre-eclampsia’, severe gestational hypertension was associated with higher risks of preterm delivery and delivery of small-for-gestational-age infants.5 Moreover, data on similarities in risk factor patterns, including immunological37 and metabolic components,38,39 point out that it may be relevant to consider these diseases as aetiologically related. The results of a common liability for these two conditions give further evidence for a common genetic aetiology.
The quantitative genetic methods used in this study have the advantage of separating the familial effects into shared environmental effects and genetic components. The shared environmental component in the model fitting analysis of pregnancy-induced hypertension was 7% and statistically significant. We were not able to further investigate which components the shared environmental effect consisted of. From previous studies, it is known that obesity, which might be a result of unhealthy familial dietary habits, is a risk factor for pre-eclampsia and gestational hypertension.40,41 It is also known that smoking might have a protective effect on the risk of developing pre-eclampsia and gestational hypertension, and familial smoking habits might be a factor in the shared environmental component.40–42
A strength in this study is that data were collected through linkage of two population-based registers, which, to our knowledge, makes it the largest study to date that has estimated the relative importance of genes and environments in the development of pre-eclampsia and gestational hypertension. The study is prospective and the diagnoses listed in the Swedish Medical Birth Register have been validated and the accuracy is reported to be good.1 However, three limitations in this study should be acknowledged. Firstly, as the possibilities to separately study pre-eclampsia and gestational hypertension improved with the introduction of ICD-9, we have only included mothers diagnosed with pre-eclampsia and gestational hypertension since 1987, the year when ICD-9 was introduced. This means that we were not able to include all women's first pregnancies. Since pre-eclampsia is known to be more prevalent in the first compared with subsequent pregnancies,14,43,44 this may partly explain the relatively low prevalence of pre-eclampsia in this study sample. The low prevalence of gestational hypertension suggests underreporting, which limits the generalisability of our findings. Secondly, previous studies have indicated that paternal genes contribute to the risk of pre-eclampsia.44,45 We have not been able to analyse the contribution of paternal genes. The effects of these genes are included in the model fitting analysis in the non-shared environment component, although we are not able to single out these effects from other non-shared environmental factors. The third consideration is that the model fitting analysis in this study does not account for the dependency between sister-pairs and therefore the CIs should be interpreted with caution.
In conclusion, we have found that genetic effects are important for the different hypertensive diseases during pregnancy. The co-morbidity between pre-eclampsia and gestational hypertension suggests a common genetic aetiology. In the continuous search for the causes of these diseases, this will add important knowledge, not only for understanding the causes of pregnancy-induced hypertensive diseases, but also for designing studies that search for susceptibility genes.