Materials and methods

The method of recruitment and the definition of pre-eclampsia used in our cohort have been published previously.4 In brief, they all had proteinuric pregnancy related hypertension as defined by the criteria of Redman and Jefferies.8 The local ethics committee approved the study and all patients gave written informed consent before taking part. At the time of this study, we had genomic DNA available from 356 women with pre-eclampsia. All the women were white northern Europeans. The 200 normotensive control samples were those reported in our earlier study4 and had no history of pregnancy induced hypertension or pre-eclampsia.

Genomic DNA was obtained from venous blood using standard methods and genotyped for the two polymorphisms using PCR/RFLP as follows.

GPIIIa C1565T POLYMORPHISM

Exon 2 of the GPIIIa gene was amplified using forward (5′ TCT GAT TGC TGG ACT TCT CTT) and reverse (5′ TCT CTC CCC ATG GCA AAG AGT) intronic primers (Genbank accession No AH00311). The 266 bp product was digested with 10 U ofNciI (New England Biolabs) overnight and fragments resolved on a 3% Metaphor^® gel (FMS Bioproducts, Rockport, USA).

GENOTYPING OF THE PTG20210A POLYMORPHISM

The 3′ UTR region of the PT gene was amplified using previously reported forward (5′ TCT AGA AAC AGT TGC CTG GC) and reverse (5′ CAA CGC CTG GTA TCA AAT GG) primers.5The 345 bp PCR product was digested overnight with 10 U ofHindIII (New England Biolabs). Fragments were again resolved using a 3% Metaphor^® gel.

The genotype and allele frequencies were compared by χ²analysis. Hardy-Weinberg equilibrium was tested using the HWE program of the LINKUTIL package (downloaded fromhttp://linkage.rockefeller.edu/ott/linkutil.htm). Odds ratios and confidence intervals were calculated by the method of Wolf.9 A p value of <0.025 (0.05/2) was considered significant to reflect a necessary Bonferroni correction.

Results

The genotyping results for the two polymorphisms are shown in table 1. Both polymorphisms were in Hardy-Weinberg equilibrium in both groups of subjects. The carrier rate for the prothrombin A20210 variant was not significantly different between our control and pre-eclamptic groups and no subjects in either group were homozygous for the A20210 variant. There was no evidence that the G20210A polymorphism was a severity marker within the pre-eclampsia group, 5/149 (3.4%) delivering before 37 weeks of gestation and 2/42 (4.8%) with severe disease (defined as having HELLP syndrome or an eclamptic seizure) were A20210 carriers. These were not significantly different frequencies from the pre-eclampsia group as a whole.

For the glycoprotein IIIa polymorphism, there was a significant excess of C98 homozygotes in the pre-eclampsia group (18/356) versus controls (1/200, p<0.01) with a corresponding excess of the C98 allele (p<0.02). Expressing the relative risk as odds ratios: for C98 versus T98 homozygosity it was 11.3 (95% CI 1.5-85.5), and for carriage of at least one C98 allele it was 1.4 (95% CI 1.0-2.1). In the pre-eclampsia group, 9/149 (6.0%) delivering before 37 weeks and 3/42 with severe disease (see above) were C98 homozygotes (7.1%). Again, these were not significantly different frequencies from the pre-eclampsia group as a whole.

Discussion

In our population, the carrier frequency for the A20210 prothrombin gene variant was within the range previously reported for white populations of 0.7-4%.10 However, our results contrast sharply with previous reports that carriage of the A20210 variant is commoner in women with pre-eclampsia.6 7 These studies found a three to four-fold increase in the frequency of this thrombophilic variant, although the numbers of subjects studied were small, consisting of just 34 subjects with pre-eclampsia compared to over 350 in our current study group.6 Differences in the severity of pre-eclampsia between studies is unlikely to be important since the carrier rate for A20210 within our own sample was not affected by severity, expressed either as premature delivery, or the presence of HELLP syndrome, or eclamptic seizure as used in the study of Kupferminc et al.6Interestingly, the most recent report from Kupferminc's group, using an increased sample size, has failed to replicate the original findings in patients with severe pre-eclampsia.12 This work is in keeping with our current negative results and suggests that the original publication may have represented a type I error. Even supposing the G20210A is a susceptibility gene variant for pre-eclampsia, its role would be restricted to certain ethnic groups; it is virtually absent, for example, in non-white populations.10

In contrast, we did see an excess of homozygotes for the T98 allele ofGPIIIa (17 in the pre-eclamptics versus 1 in the controls, table 1) and overall the frequency of the T98 allele was significantly increased in our pre-eclamptic subjects (20.1% versus 14%, p<0.01). This raises the intriguing possibility that theGPIIIa C98 polymorphism may predispose women to pre-eclampsia. Previous association studies with this polymorphism have modelled its effects as a dominant allele, largely reflecting the bias of early studies that could type on the basis of the presence or absence of the Pl(A)1 and Pl(A) 2 surface antigens. Expressing our data in this fashion gives an odds ratio for women carrying at least one T98 allele of 1.4. Although this appears a rather modest effect, it is similar to that previously published for the factor V Leiden and MTHFR variants in women with pre-eclampsia.2 3

The GPIIIa gene codes for the β3 subunit of the β3 integrin subfamily; this consist of αIIbβ3 (glycoprotein IIb/IIIa) and αvβ3. The glycoprotein IIb/IIIa receptor is confined to platelets and megakaryocytes and has a key role in platelet aggregation by binding fibrinogen and von Willebrand factor. In contrast, the αvβ3 integrin is a widely expressed receptor for ligands such as vitronectin and fibronectin. It appears to be an important player in tumour angiogenesis, but is also expressed by invading trophoblast suggesting a role in placentation.13Indeed, mice knockouts in which the β3 integrin is deficient have normal implantation but develop a characteristic placental defect.14 The β3 integrin has also been implicated in the failure of the cytotrophoblast to adopt a vascular phenotype in pre-eclampsia.15

Previous epidemiological work on the C98TGPIIIa polymorphism has focused on its role as a risk allele for arterial thrombosis, specifically for acute coronary syndromes. Weiss et al 16 initially reported that carriage of the T98 variant of GPIIIa was a risk factor for myocardial infarction in young patients. This has been confirmed in a large cohort of sib pairs with premature heart disease and probably explains why some studies have failed to replicate the original work of Weiss et al 16 because they used elderly patient cohorts.17 Clear evidence now exists that the T98 GPIIIa allele that carries the thrombotic risk is functionally distinct from the C98 wild type receptor. This is apparent both in terms of the level of IIb/IIIa expression on activated platelets as well as in the adhesion properties of cells expressing αvβ3.18 19 Given the importance of platelet and trophoblast function in the pathogenesis of pre-eclampsia, the GPIIIa gene is a very attractive biological candidate gene for pre-eclampsia and the functional C98T polymorphism in particular. Our findings here that the thrombotic risk variant, T98, is commoner in pre-eclamptic women suggests that we may have identified a new risk allele for the disease.

In summary, we have not been able to show any association of the prothrombin polymorphism, G20210, with pre-eclampsia. There is significant excess, however, of the thrombotic variant of the β3 integrin protein GPIIIa in the same population. This result needs confirmation in other populations, preferably using robust family based association studies, but it suggests that integrin polymorphisms might be an important new focus for studies on genetic risk factors for pre-eclampsia.