A study of three candidate genes for pre-eclampsia in a Sinhalese population from Sri Lanka

Abstract

Aim:  The aim of these investigations was to study three candidate genes for pre-eclampsia – epidermal growth factor (EGF), transforming growth factor alpha, and angiotensinogen – in pregnant Sinhalese women from Sri Lanka, the first such study undertaken in this ethnic group. Reproducibility of results of genetic association studies of candidate genes for pre-eclampsia has not been consistent across populations. One of the factors that may contribute to such inconsistencies is genetic stratification due to population admixture. We therefore compared the allele frequencies of these candidate genes in healthy Sri Lankan subjects from three ethnic groups – Sinhalese, Sri Lankan Tamils and Moors – and in white Western Europeans.

Methods:  Allele frequencies were established in 80 subjects from each of four populations (Sinhalese, Sri Lankan Tamils, and Moors in Sri Lanka and white Western Europeans in the UK). A further 175 Sinhalese women with pre-eclampsia and 171 normotensive Sinhalese controls were genotyped at eight single nucleotide polymorphisms in the candidate genes.

Results and Discussion:  In all genes haplotype and allele frequencies were comparable within the three Sri Lankan populations, but differed significantly from those in the white Western European population. Consequently cryptic population stratification is unlikely to have significant effects on allele or haplotype frequencies of the genes examined in this case–control study of Sinhalese women which showed a marginal association for EGF haplotypes and genotypes with pre-eclampsia (P = 0.031). This association requires replication in other populations.

Introduction

Pre-eclampsia is a life-threatening disorder of pregnancy recognized clinically by the onset of hypertension and proteinuria.¹ It is estimated that annually 40 000 mothers and 900 000 babies lose their lives as a result of the hypertensive disorders of pregnancy, and the bulk of this burden is borne by the developing world.² Pre-eclampsia is 3- to 5-fold more common in first degree relatives of affected women than in the general population.³ Twin studies report pairwise concordance rates of 0.25 and 0.06 for pre-eclampsia among monozygotic and dizygotic twin pairs respectively and estimates heritability (influence of maternal genes) and the non-shared environmental effect (including influence of fetal/paternal genes) at 0.54 and 0.46, respectively.⁴ A growing number of candidate genes have been tested in Western populations,⁵ but there are relatively few from developing countries, and none from Sri Lanka, where the incidence of pre-eclampsia is high [5% of pregnancies in one hospital-based survey⁶]. We undertook a case–control genetic marker disease association study of candidate genes in Sinhalese women in Colombo, Sri Lanka.

Pre-eclampsia is characterized by impaired invasion of trophoblast into maternal tissues at the site of placentation,⁷ a process which is regulated by growth factors, cell adhesion molecules, cytokines and proteinases.⁸ Genes encoding growth factors at the maternal-fetal interface are therefore candidates for pre-eclampsia. Three such genes, epidermal growth factor (EGF), transforming growth factor α (TGFA) and angiotensinogen (AGT) were selected for this investigation.

EGF and TGFA are novel candidates which have not been tested in other populations. They have been identified at the maternal-fetal interface, and both are ligands for epidermal growth factor receptors, which are expressed by the trophoblast.⁹ Furthermore, EGF maps to chromosome 4q25–27,¹⁰ approximately 57cM away from the markers bordering the suggestive locus for pre-eclampsia detected in a genome-wide scan of Finnish families;¹¹ and TGFA maps to chromosome 2p13¹² within the boundary of the most likely location for a pre-eclampsia susceptibility gene identified in genome-wide scans of Icelandic and Australasian families.^13,14 Two single nucleotide polymorphisms (SNP) that have possible functional effects have been described in the EGF gene. These are NM_001963:c.61G>A, which has been shown to have an effect on the production of EGF in peripheral blood mononuclear cells (PBMC)¹⁵ and NM_001963:c.2566G>A that occurs in exon 14 of the gene and results in an amino acid substitution in the low density lipoprotein receptor homology domain of EGF.¹⁶ There is also a cluster of SNP at nucleotides 3822, 3827, and 3851 in the 3′UTR of the TGFA gene (AF123243:g.599G>A; AF123243:g.604T>C; AF123243:g.628T>C respectively). These three SNP define four haplotypes. They have been found to be associated with the development of disorders involving tissue remodeling.^17,18 Polymorphisms in 3′UTR conserved regions of genes can play a role in messenger ribonucleic acid stability or tissue-specific targeting.¹⁹

AGT encodes angiotensinogen, the precursor of angiotensin II, which is a potent vasoconstrictor, a major determinant of salt and water homeostasis and a growth factor. AGT is mainly synthesized in the liver, but it is expressed in a wide range of tissues, including the placenta. AGT has been implicated in previous studies of pre-eclampsia and essential hypertension^20,21 and angiotensin II AT1 receptor expression is increased fivefold in pregnancies complicated by pre-eclampsia.²² Three single nucleotide polymorphisms, AY436323:g.6679C>T and AY436323:g.6862T>C in exon 2, and AY436323:g.14321C>A in the 3′ UTR, define four common haplotypes in AGT which are associated with differential effects on plasma angiotensinogen levels and blood pressure.^{23, 24}

Case–control genetic marker disease association studies in complex disorders such as pre-eclampsia have generated many positive results which have not been consistently replicated in subsequent studies.⁵ Possible explanations for the lack of consistency include phenotypic variation between studies, population stratification, lack of power of many small scale studies in the face of expected modest genetic effects, and differences in genetic background of different populations studied. We addressed the problem of phenotypic variation by rigorous application of an international consensus definition of pre-eclampsia¹ at the time of recruitment of the case and control groups, and by restricting recruitment to nulliparous women. To assess the effects of unsuspected population stratification we undertook a comparison of allele/haplotype frequencies in healthy subjects from the three predominant ethnic groups in Sri Lanka: Sinhalese, Sri Lankan Tamils and Moors; and white Western Europeans in Nottingham, UK.

The objectives of these investigations therefore, were (i) to establish allele and haplotype distributions of polymorphisms in three candidate genes for pre-eclampsia – EGF, TGFA and AGT – and measures of linkage disequilibrium, in three major ethnic groups in Sri Lanka, and to compare that with white Western Europeans in the UK; and (ii) to conduct a case–control genetic marker disease association study of the three candidate genes in nulliparous Sinhalese women with and without pre-eclampsia.

Materials and Methods

Subjects

Sri Lankan subjects were recruited for these studies in Colombo, Sri Lanka between August 2001 and January 2003 according to protocols approved by the Ethics Review Committee, Faculty of Medicine, University of Colombo, Sri Lanka. All subjects gave written informed consent to participate and the study was carried out in accordance with the Declaration of Helsinki (2000) of the World Medical Association.

Women with pre-eclampsia were recruited at the time of diagnosis, and controls were recruited within 24 hours of a normal full term delivery. Subjects for the population genetic study were recruited from among women attending antenatal clinics, staff and students of the University of Colombo, and persons accompanying patients attending general practice clinics in Colombo. Samples were obtained from 80 (50% male) volunteers each from the three main racial groups in Sri Lanka – Sinhalese, Sri Lankan Tamils, and Moors. The mean ages of these three groups of volunteers were 33 years (range 17 to 59), 28 years (range 18 to 45), and 27 years (range 17 to 45), respectively. The ethnicity of subjects was established by grandparental ethnicity.

The 80 (50% male) white Western European samples for population studies were from an existing DNA resource at the University of Nottingham. This collection had been made from blood donated by volunteers from among healthy blood donors in Nottingham, mean age 37 (range 19 to 62), according to a protocol approved by the Hospital Ethics Committee, Queens Medical Centre, Nottingham, UK.

Sinhalese women for the association study of pre-eclampsia were recruited from among pregnant women in two tertiary care hospitals in Colombo. Pre-eclampsia was defined as hypertension of ≥140/90 mm Hg on two occasions six hours apart occurring after 20 weeks of gestation in a previously normotensive woman together with proteinuria of ≥1+ on the urine protein heat coagulation test (HCT; equivalent to 500 mg/day)²⁵ not associated with urinary tract infection or ruptured membranes. The use of the HCT as a measure of proteinuria has been validated in these maternity units.²⁵ A group of normotensive pregnant Sinhalese women were also recruited as controls. Exclusion criteria applied were: non-Sinhalese or women of mixed ethnicity, current pregnancy fathered by a non-Sinhalese man; renal disease, chronic hypertension, or persistent proteinuria (defined as ≥1+ on the HCT in the first three urine samples tested in pregnancy with or without urinary tract infection); ischemic heart disease, cerebrovascular accidents, or insulin or non-insulin dependent diabetes mellitus; body mass index (BMI) ≥30 kg/m² based on height and weight measured at the antenatal booking visit, or postpartum in the case of unbooked pregnancies (to minimize phenotypic heterogeneity, as obesity is itself a risk factor for pre-eclampsia); a previous abortion after 12 weeks of gestation (because the uterus mimics a parous uterus due to uterine vascular re-modeling in such cases); hydatidiform mole, multiple gestations, or gestational diabetes in the current pregnancy. The initial case–control collection had 180 cases and 180 controls that were matched for ethnicity, age and BMI. A detailed discussion of phenotypic features and recruitment of this cohort has already been published.²⁶ Four women with pre-eclampsia had elevated blood pressure at follow up at 6 weeks. One woman with pre-eclampsia died. Nine control women had delivered babies who had a corrected birthweight centile of <5, compatible with a diagnosis of fetal growth restriction which may share a common pathology with pre-eclampsia.⁷ As such these five cases and nine controls were excluded from the current analysis leaving 175 case subjects and 171 control subjects. A summary of the phenotypic features of these subjects are shown in Table 1. The cases and controls did not differ significantly in age or BMI.

Table 1.  Clinical features of Sinhalese women with pre-eclampsia (PE) and controls

		PE (n = 175)		Control (n = 171)		P
† The control women had blood pressures of less than 140/90 mm Hg throughout the pregnancy. SD, standard deviation. M(iqr), median (inter quartile range). Significance (P) is for comparison between women with pre-eclampsia and control women.
Mean age (years)	(SD)	27.0	(5.3)	27.1	(5.1)	0.806
Mean body mass index (kg/m^2)	(SD)	21.0	(3.3)	20.7	(3.2)	0.463
Maximum systolic blood pressure (mm Hg)†	M(iqr)	160	(158, 175)
Maximum diastolic blood pressure (mm Hg)†	M(iqr)	110	(110, 120)
Gestation at delivery (weeks)	M(iqr)	35	(32, 38)	39	(38, 40)	<0.001
Corrected birthweight centile	M(iqr)	5.7	(0.4, 33)	22.3	(48, 76)	<0.001
Sex of baby	Male (%)	102	(58.3)	105	(61.4)	0.629

Genotyping

DNA was extracted from venous blood using QiaAmp blood mini DNA extraction kits (Qiagen). The above mentioned SNP, all of which have a minor allele frequency of >5%, were selected for genotyping based on their reported functional effects and/or value in defining haplotype diversity. Genotyping was undertaken using polymerase chain reaction (PCR) followed by restriction digestion [EGF – NM_001963:c.61G>A; TGFA – AF123243:g.604T>C, AF123243:g.628T>C¹⁵] or allele specific oligonucleotide (ASO) hybridization [TGFA – AF123243:g.599G>A, AGT – AY436323:g.6679C>T, AY436323:g.6862T>C, AY436323:g.14321C>A²⁷], or by mutagenically separated (MS) PCR [EGF – NM_001963:c.2566G>A²⁸]. Full details of genotyping methods of TGFA polymorphisms, including oligonucleotide primer and probe sequences and PCR conditions, are available on request. As quality assurance measures, sequenced positive control samples and DNA-free blanks were included in all batches, and all genotypes were confirmed by two independent observers.

Statistical Analysis

The χ² test was used to test the genotypes at each polymorphic locus for Hardy-Weinberg equilibrium (HWE) in the three populations and in cases and controls separately; and to compare allele frequency differences between cases and controls. The COCAPHASE program of the UNPHASED software suite ²⁹ was used to estimate haplotype frequencies, to measure pairwise linkage disequilibrium (D′ and r²), and for hypothesis testing. Rare haplotypes occurring at frequencies of ≤0.05 were dropped from the analysis in both case and control groups unless otherwise stated. QUANTO version 0.5 was used for power calculations.³⁰

Results

The results of the population genetic study are shown in Tables 2 and 3 giving haplotype frequency estimates of the EGF, TGFA, and AGT genes and measures of linkage disequilibrium, respectively. All polymorphisms in each population were in HWE. The haplotype frequencies of the three Sri Lankan ethnic groups were similar; the single exception was a significant difference in TGFA haplotype frequencies between Sinhalese and Moors (P = 0.007). By contrast haplotype frequencies in all genes differed significantly between white Western Europeans and some or all Sri Lankan ethnic groups.

Table 2.  Haplotypes defined by single nucleotide polymorphisms (SNP) in the candidate genes and variant allele frequencies of SNP in the population

Genes	Haplotypes/variant alleles			Haplotype/variant allele frequency				Significance (P)
				Sinhalese (S)	Tamils (T)	Moors (M)	Whites (W)	S v T	S v M	T v M	S v W	T v W	M v W
* P > 0.05. The P-values refer to the comparison of the haplotypes between the different populations. The reference sequences are as follows: EGF – NM_001963; TGFA – AF123243; AGT – AY436323. AGT, angiotensinogen; EGF, epidermal growth factor; TGFA, transforming growth factor α.
EGF	c.61	c.2566
	G	G		0.06	0.06	0.07	0.12	*	*	*	0.012	*	*
	G	A		0.53	0.47	0.44	0.37
	A	G		0.36	0.42	0.45	0.48
	A	A		0.05	0.04	0.05	0.03
	c.61A			0.41	0.46	0.50	0.51
	c.2566A			0.58	0.51	0.48	0.40
TGFA	g.599	g.604	g.628
	G	T	T	0.20	0.24	0.23	0.21	*	0.007	*	*	0.002	<0.001
	G	C	T	0.56	0.48	0.51	0.62
	G	C	C	0.10	0.08	0.02	0.11
	A	C	T	0.14	0.20	0.23	0.06
	g.599A			0.14	0.20	0.23	0.06
	g.604T			0.20	0.24	0.23	0.21
	g.628C			0.10	0.08	0.02	0.11
AGT	g.6679	g.6862	g.14321
	C	T	A	0.31	0.25	0.24	0.29	*	*	*	<0.001	<0.001	<0.001
	C	T	C	0.06	0.11	0.13	0.34
	C	C	A	0.02	0.01	0.01	0.04
	C	C	C	0.53	0.51	0.50	0.22
	T	C	C	0.08	0.11	0.11	0.09
	g.6679T			0.09	0.11	0.11	0.11
	g.6862C			0.64	0.63	0.63	0.37
	g.14321A			0.33	0.27	0.25	0.34

Table 3.  Linkage disequilibrium analysis

Gene	Loci combination		Sinhalese		Tamils		Moors		Whites
			D′	r^2	D′	r^2	D′	r^2	D′	r^2
Lewontin's disequilibrium coefficient (D′) and correlation coefficient (r^2). The reference sequences are as follows: EGF – NM_001963; TGFA – AF123243; AGT – AY436323. AGT, angiotensinogen; EGF, epidermal growth factor; TGFA, transforming growth factor α.
EGF	c.61G>A	c.2566G>A	0.78	0.59	0.83	0.63	0.78	0.58	0.86	0.52
TGFA	g.599G>A	g.604T>C	1.00	0.04	1.00	0.08	1.00	0.09	1.00	0.02
	g.599G>A	g.628T>C	0.38	0.002	0.12	<0.01	0.05	<0.01	0.004	<0.01
	g.604T>C	g.628T>C	1.00	0.03	1.00	0.03	1.00	0.008	1.00	0.03
AGT	g.6679C>T	g.6862C>T	1.00	0.05	1.00	0.07	1.00	0.08	1.00	0.21
	g.6679C>T	g.14321C>A	1.00	0.05	1.00	0.05	1.00	0.04	0.58	0.02
	g.6862C>T	g.14321C>A	0.90	0.69	0.92	0.53	0.91	0.47	0.60	0.11

The results of haplotype analysis and single locus genotype analysis in Sinhalese women with pre-eclampsia and normotensive controls are shown in Tables 4 and 5, respectively. All polymorphisms in cases and controls were in HWE. Of the three genes tested, only EGF haplotypes demonstrated any evidence of association with pre-eclampsia, and this was of marginal significance (P = 0.031). This effect was mainly due to an excess of the NM_001963:c.2566G allele and the NM_001963:c.61G; c.2566G haplotype in women with pre-eclampsia.

Table 4.  Haplotype analysis of case–control data

Genes	Haplotypes			Haplotype Frequency
				PE (n = 175)	Controls (n = 171)	P
Three rare haplotypes that have a total pooled frequency of 0.015 in both case and control groups were excluded from the angiotensinogen (AGT) case–control analysis. There were no rare haplotypes in the epidermal growth factor (EGF) or transforming growth factor α (TGFA) genes. The reference sequences are as follows: EGF – NM_001963; TGFA – AF123243; AGT – AY436323. PE, women with pre-eclampsia.
EGF	c.61	c.2566
	G	G		0.12	0.06	0.031
	G	A		0.42	0.39
	A	G		0.43	0.49
	A	A		0.04	0.06
TGFA	g.599	g.604	g.628
	G	T	T	0.18	0.20	0.716
	G	C	T	0.59	0.58
	G	C	C	0.07	0.06
	A	C	T	0.16	0.16
AGT	g.6679	g.6862	g.14321
	C	T	A	0.31	0.29	0.673
	C	T	C	0.09	0.09
	C	C	C	0.50	0.48
	T	C	C	0.09	0.12

Table 5.  Results of analysis of genotypes of the cases and controls (variant allele frequency of each single nucleotide polymorphism is also given in the table)

Discussion

In these investigations we set out to conduct studies of susceptibility genes for pre-eclampsia using SNP and the haplotypes defined by these SNP as genetic markers. We examined the haplotype distribution in four racial groups to study how the background genetic make-up of the population could affect such studies. These results show that the three Sri Lankan populations are homogeneous with regard to the majority of SNP used in these investigations; but differed significantly from those in the white Western European population. Consequently cryptic population stratification is unlikely to have significant effects on allele or haplotype frequencies of the genes examined in the case–control study of Sinhalese women. However, the significant differences in allele frequencies of the Sri Lankan populations compared to white western Europeans, especially with regard to the EGF gene polymorphisms as confirmed by allele frequencies for white western Europeans reported in previous studies,^15,16 have implications if an attempt is made to replicate these findings in that population.

Another factor that could have influenced our case–control study results is sample size and power. The sample size of this case–control study had 80% power to detect a genotype relative risk of between 1.8 and 2.7, depending on allele frequencies, at a significance level of 0.05. It would be underpowered to detect lower risks associated with these candidate alleles, especially if a lower threshold of significance was applied to correct for multiple hypothesis testing. It is interesting to note in this context the negative finding of the GOPEC consortium with respect to the association of AGT gene polymorphisms with pre-eclampsia in a collection 657 white Western European child–parent trios that had adequate size and power to detect such an association.³¹

In this study an association of marginal significance was found for the association of the EGF gene NM_001963:c.2566G allele and NM_001963:c.61G; 2566G haplotype with pre-eclampsia in Sinhalese women. The high D′ yet small r² values between these two SNP suggest that the significant association of c.2566G might also be accounted for by c.61G as supported by results of haplotype analysis. The marginal differences in allele frequencies however, did not justify further refinement of the analysis. Since multiple hypotheses were tested, this association would not have been significant if a more stringent cut-off for significance correcting for such multiple testing was applied. Interestingly, however, we recently reported the association of the maternal EGF c.2566G allele with a higher weight at birth than the c.2566A allele in women delivering healthy babies after uncomplicated pregnancies in three populations, including the control population described in this paper.²⁸ In addition, in pregnancies in white Western European women affected by fetal growth restriction the c.2566A allele was shown to be preferentially transmitted to the affected babies by their parents. These observations suggest that the c.2566A allele is associated with low birthweight in both healthy and growth-restricted pregnancies. The association of the c.2566G allele with pre-eclampsia, a disorder with a well-established association with fetal growth restriction, is therefore a counterintuitive finding. It would be interesting to investigate EGF allele transmission in babies affected by pre-eclampsia in a suitable sample collection to develop a more comprehensive understanding of the interaction at the maternal–fetal interface at the time of placentation. Studies of the effect of these polymorphisms on EGF expression in trophoblasts would further add to our understanding of this interaction.

In this study, we sought to minimize phenotypic heterogeneity by the application of rigorous inclusion and exclusion criteria. Nevertheless, there is some evidence that severe, early onset disease presenting before 34 weeks gestation represents a more homogeneous subgroup. Only 77 of the women in our study had early onset disease, too few to justify subgroup analysis. Targeted recruitment of a larger number of these severe cases for genetic analysis, or metanalysis of this subgroup from a number of large studies, however, may prove to be an informative approach.

In conclusion, the association of marginal significance of pre-eclampsia with EGF haplotype detected in Sinhalese women requires verification in further independent studies using a sufficiently powered sample collection. The EGF allele and haplotype frequency data reported here would be useful in planning such a study.

Acknowledgement

We wish to thank all volunteers and the consultant obstetricians and the staff of the De Soysa Hospital for Women and the Castle Street Hospitals for Women in Colombo, Sri Lanka for their support. V. H. W. D. held a PhD studentship funded jointly by the President's Fund Sri Lanka, University of Colombo, Sri Lanka and the University of Nottingham, UK. The authors declare that they have no conflicts of interest.