Objective
This study investigates the association of allelic variation in angiogenic pathway genes and preeclampsia.
Study Design
Data for cases with preeclampsia and term control subjects were collected prospectively. Maternal DNA was extracted, and 124 tagging single nucleotide polymorphisms in 6 genes (vascular endothelial growth factor A, B, and C; fms-like tyrosine kinase 1 and 4; endoglin) were genotyped. Multivariable logistic regression was used to evaluate the association between tagging single nucleotide polymorphisms and preeclampsia; data were controlled for age. All models were evaluated in black women and white women separately. Haplotype analyses were performed.
Results
We analyzed data from 606 women (489 black women [184 cases] and 117 white women [32 cases]). In black women, the fms-like tyrosine kinase 1 rs12584067 (odds ratio [OR], 1.55; 95% confidence interval [CI], 1.01–2.36; P = .05) and rs7335588 (OR, 1.61; 95% CI, 1.06–2.43; P = .01) and the vascular endothelial growth factor C rs1485766 (OR, 1.56; 95% CI, 1.05–2.30; P = .03) and rs6838834 (OR, 1.60; 95% CI, 1.05–2.45; P = .03) single nucleotide polymorphisms were associated with preeclampsia. In white women, the fms-like tyrosine kinase 1 rs722503 (OR, 2.12; 95% CI, 1.07–4.19; P = .03), fms-like tyrosine kinase 4 rs307826 (OR, 3.06; 95% CI, 1.18–7.91; P = .01), and vascular endothelial growth factor C rs7664413 (OR, 2.04; 95% CI, 0.99–4.17; P = .04) single nucleotide polymorphisms were associated with preeclampsia.
Conclusion
Allelic variations in the fms-like tyrosine kinase 1 and vascular endothelial growth factor C genes are associated with preeclampsia in both ethnic groups.
Preeclampsia is a pregnancy-specific syndrome that is defined broadly as the presence of hypertension and proteinuria. It affects approximately 5-7% of all pregnancies. However, it encompasses a wide spectrum of diseases that range from simple hypertension to the development of seizures and multiorgan system involvement. It is a major contributor to maternal and perinatal morbidity worldwide and accounts for approximately 20% of maternal deaths in the United States.
The inability to develop satisfactory prevention and prediction strategies for preeclampsia is not due to lack of concerted research effort. A variety of theories have been proposed to explain the development of preeclampsia. Epidemiologic studies have demonstrated consistent risk factors for the development of preeclampsia that have included chronic hypertension, obesity, pregestational diabetes mellitus, collagen vascular disorders, renal insufficiency, and maternal race. Because observable maternal phenotypes alone have not been able to explain most cases of preeclampsia, there has been a shift in focus to the evaluation of genetic predispositions to this disease, as has occurred in research of other complex disorders. Suggesting a genetic component to preeclampsia are (1) studies that demonstrate an increased risk of recurrence of preeclampsia in subsequent pregnancies, (2) studies that observe an increased risk in first-degree relatives of women with preeclampsia, and (3) familial studies that demonstrate that a family history that is positive for cardiovascular disease also predicts the occurrence of preeclampsia.
Investigation into the genetic cause of preeclampsia has focused largely on candidate genes that are related to thrombophilias, folate metabolism, lipid metabolism, oxidative stress, and the renin-angiotensin system. To date, these genetic studies have not yielded consistent results, and investigation of additional candidate gene pathways is warranted. Soluble fms-like tyrosine kinase (sFlt1) is a soluble deactivating receptor for vascular endothelial growth factor (VEGF) and placental growth factor. Both VEGF and placental growth factor are potent stimulators of angiogenesis and regulators of endothelial function. It has been demonstrated that increased levels of sFlt1 are associated with preeclampsia and are present in the serum of patients who are destined to have preeclampsia as early as 5 weeks before the onset of clinical symptoms. This suggests a role for this biomarker in the cause of preeclampsia rather than this elevated level being a consequence of the disease. Levels of endoglin, which is a glycoprotein that is involved in vascular remodeling, have also been elevated in the serum of preeclamptic women compared with control subjects, with significant elevations detectable before the onset of clinical symptoms of preeclampsia. Given these findings, genes that are involved in the angiogenesis pathway are suited ideally for exploration in understanding the cause of preeclampsia.
This study aimed to investigate whether haplotype-tagging single-nucleotide polymorphisms (tagSNPs) in 6 genes in the angiogenic pathway (VEGFA, -B, -C; FLT1 and -4, and endoglin) are associated with preeclampsia in white and black women. To the best of our knowledge, no other study has investigated the role of tagSNPs in these biologically plausible genes in this number of women with preeclampsia.
Materials and Methods
Study population
We performed a prospective case-control study (Preeclampsia: Mechanisms and Consequences) between March 2005 and August 2007 at the Hospital of the University of Pennsylvania. Institutional Review Board approval was obtained before enrollment began. Cases were women with preeclampsia; control subjects were women who delivered at term (≥37 weeks of gestation) without preeclampsia. All women who were admitted to Labor and Delivery with preeclampsia were eligible for enrollment and were invited to participate in the overall study. Cases were identified on the basis of prespecified maternal criteria for preeclampsia, according to standard criteria of the American College of Obstetricians and Gynecologists. Preeclampsia was defined as elevated blood pressure (≥140/90 mm Hg on 2 measurements ≥6 hours apart) with ≥1+ proteinuria. Based on these prespecified criteria, case eligibility was determined at the time of enrollment and not by the treating physician. Control subjects were enrolled prospectively from women who underwent delivery at term (≥37 weeks of gestation) for scheduled induction of labor, scheduled cesarean section delivery, spontaneous rupture of membranes, or term labor. Race and ethnicity were categorized according to the standard National Institutes of Health definitions and were obtained by direct patient self-report at the time of enrollment. Cases and control subjects in the overall study were frequency matched for self-reported race (black vs other). However, only patients who identified as black or white are included in this analysis. Additional descriptions of this study are provided elsewhere.
Trained research nurses collected information on height, race, ethnicity, and history of hypertension by patient interview at the time of enrollment. Additional medical history was collected from prenatal and hospital chart abstraction by trained research nurses. The primary investigator (S.K.S.) reviewed all data sheets for completeness, consistency, and accuracy.
Maternal whole blood was collected from a subset of subjects who were enrolled from 2006-2007 within 24 hours of enrollment. After centrifugation, serum samples were aliquoted and placed at –80°C. Commercially available enzyme-linked immunosorbent assays were used to measure levels of sFlt1 (nanograms per milliliter) in the maternal serum (R &D Systems, Minneapolis, MN) by personnel who were neither aware of the case-control status nor genotyping results.
Genotyping methods
Maternal DNA was derived from whole blood samples that were collected from all women within 24 hours of enrollment. Specimens were stored and then processed within approximately 6 hours of collection. Samples were genotyped for the IBCv2 array (Illumina Inc, San Diego, CA) that contained 50,000 gene-centric SNPs that were selected for their relevance to cardiovascular, metabolic, and inflammatory syndromes. From the panel of genotyped SNPs, this study focused on 124 tagSNPs in 6 genes that were involved in angiogenesis. Thirty-three SNPs were monomorphic in white women, which resulted in a total of 91 SNPs that were available for analysis in this ethnic group. Three SNPs were monomorphic in black women, which resulted in a total of 121 SNPs that were available in this ethnic group.
Statistical analyses
Within each ethnic group, demographic characteristics were compared between preeclampsia cases and control subjects. For continuous variables, the means were calculated and compared with the use of a Student t test. χ 2 analyses or Fisher’s exact tests were used to compare categoric variables, as appropriate. Agreement of genotype frequencies with Hardy-Weinberg equilibrium expectations was evaluated by a χ 2 goodness-of-fit test. SNPs that deviated from Hardy-Weinberg equilibrium ( P < .001) in control subjects were not analyzed. Logistic regression was used to evaluate the association between the angiogenic factor gene SNPs and preeclampsia; all models were adjusted for maternal age. Odds ratios were calculated with an assumption of an additive effect of the SNP alleles. All models were evaluated separately in white and black women. To account for the number of statistical tests that were performed in each ethnic group, a permutation-based procedure that involved 1000 null permutations was performed to generate empiric probability values.
For genes that demonstrated a significant association with preeclampsia for at least 1 SNP, haplotypes were estimated in the program PLINK with the use of the expectation-maximization algorithm. Logistic regression was used for the haplotype-based association analyses. For H haplotypes, a single H-1 degree of freedom omnibus test was performed that jointly evaluated all haplotype effects. The omnibus test was applied to haplotypes estimated within a sliding window of fixed haplotype size of 2-4 SNPs with the use of the program PLINK. The program Graphical Assessment of Sliding P values (GrASP) was used to visualize probability values from the haplotype analyses and to help identify gene regions of interest.
Results
We analyzed data from 606 women; 489 black women (305 cases) and 117 white women (32 cases) were evaluated. The demographic characteristics of our population are shown in Table 1 . Probability values that are reported represent within-race group comparison (black cases vs control subjects and white cases vs control subjects). Of the 91 SNPs that were available for analysis in white women, all were in accordance with Hardy-Weinberg expectations. Of the 121 SNPs in black women, 120 were in accordance with Hardy-Weinberg expectations.
| Variable | Black cases (n = 184) | Black control subjects (n = 305) | P value | White cases (n = 32) | White control subjects (n = 85) | P value |
|---|---|---|---|---|---|---|
| Mean maternal age, y | 26.2 ± 6.2 | 25.5 ± 6 | .22 | 33.4 ± 5.7 | 32.3 ± 5.2 | .35 |
| Mean body mass index at screening, kg/m 2 | 30.9 ± 9.8 | 29.9 ± 8 | .26 | 26.8 ± 6 | 25.4 ± 5.8 | .25 |
| Chronic hypertension, n (%) | 27 (14.7) | 16 (5.3) | < .001 | 5 (15.6) | 2 (2.4) | .007 |
| Pregestational diabetes mellitus, n (%) | 6 (3.3) | 8 (2.6) | .78 | 1 (3.1) | 2 (2.4) | 1.00 |
| Primiparous, n (%) | 87 (47.3) | 96 (31.5) | < .001 | 21 (65.6) | 39 (45.9) | .06 |
| History of preeclampsia, n (%) | 41 (22.3) | 27 (8.9) | < .001 | 5 (15.6) | 2 (2.4) | .002 |
| Mean gestational age at delivery, wk | 35.8 ± 3.8 | 39.1 ± 1.3 | < .001 | 35.6 ± 3.4 | 39 ± 1.2 | < .001 |
| Intrauterine grown restriction <10%, n (%) | 31 (57) | 53 (17) | < .001 | 9 (28) | 4 (4.7) | < .001 |
The risk allele is defined as the allele that conferred increased risk of preeclampsia in a particular ethnic group. The frequency of the risk allele is reported based on the frequency in control subjects. Results are reported with the use of empiric probability values. In black women, the FLT1 rs12584067 (odds ratio [OR], 1.55; 95% confidence interval [CI], 1.01–2.36; P = .05) and rs7335588 (OR, 1.61; 95% CI, 1.06–2.43; P = .01) SNPs are associated with preeclampsia and are not in linkage disequilibrium (r 2 = 0.05). Two VEGFC SNPs, rs1485766 (OR, 1.56; 95% CI, 1.05–2.30; P = .03) and rs6838834 (OR, 1.60; 95% CI, 1.05–2.45; P = .03) SNPs, were also associated significantly with preeclampsia and not in linkage disequilibrium (r 2 = 0.02). A single VEGFA SNP rs833068 approached significance (OR, 1.28; 95% CI, 0.99–1.67; P = .05). No VEGFB, endoglin, or FLT4 SNPs were associated with preeclampsia in black women. In white women, FLT1 rs722503 (OR, 2.12; 95% CI, 1.07–4.19; P = .03), FLT4 rs307826 (OR, 3.06; 95% CI, 1.18–7.91; P = .01), and VEGFC rs7664413 (OR, 2.04; 95% CI, 0.99–4.17; P = .04) SNPs were associated significantly with preeclamspia. No VEGFA, VEGFB, or endoglin SNPs were associated with preeclampsia in white women ( Table 2 ).
| Gene | Single nucleotide polymorphism identification no. | Risk allele | Risk allele frequency | Odds ratio (95% CI) | P value | Empiric P value |
|---|---|---|---|---|---|---|
| Black women | ||||||
| Endoglin | rs17557600 | T | 0.90 | 1.46 (0.98-2.17) | .06 | .057 |
| FLT1 | rs12584067 | G | 0.14 | 1.55 (1.01–2.36) | .04 | .047 |
| FLT1 | rs670084 | T | 0.38 | 1.29 (0.97–1.71) | .08 | .069 |
| FLT1 | rs7335588 | G | 0.93 | 1.61 (1.06–2.43) | .02 | .013 |
| FLT1 | rs880704 | G | 0.92 | 1.47 (0.94–2.31) | .09 | .087 |
| FLT1 | rs9319425 | C | 0.45 | 1.26 (0.98–1.62) | .07 | .07 |
| VEGFA | rs833068 | A | 0.57 | 1.28 (0.99–1.67) | .06 | .053 |
| VEGFC | rs1485766 | A | 0.90 | 1.56 (1.05–2.30) | .03 | .026 |
| VEGFC | rs6838834 | C | 0.91 | 1.60 (1.05–2.45) | .03 | .032 |
| White women | ||||||
| FLT1 | rs722503 | A | 0.79 | 2.12 (1.07–4.19) | .03 | .034 |
| FLT4 | rs307826 | A | 0.94 | 3.06 (1.18–7.91) | .02 | .007 |
| VEGFC | rs7664413 | C | 0.85 | 2.04 (0.99–4.17) | .05 | .036 |
Given the significant univariate association with SNPs in FLT1 and VEGFC in black and white women, these genes were selected for haplotype estimation and association with preeclampsia. Although a single SNP in FLT4 was associated significantly with preeclampsia in white women, this was the only tagSNP to be evaluated in white women; therefore, haplotypes were not estimated for FLT4 in white women. The results from an omnibus test that was evaluated for sliding windows of haplotypes comprised of 2-4 SNPs in each window is shown for FLT1 in Figure 1 and VEGFC in Figure 2 . This novel haplotyping approach identified 2 regions of significantly associated SNPs in the FLT1 gene in black women that were not identified by univariate analyses alone. In black women, an omnibus test of haplotypes in FLT1 comprised of 3 SNPs (rs7999615, rs7992940, and rs10507386) was associated significantly with preeclampsia ( P = .033). This region encompassed 27919011 base pairs (bp) to 27926554 bp and corresponds to an intronic gene region. An additional intronic region between 27817330 bp and 27822832 bp was identified in black women by analysis of 3 FLT1 SNPs (rs3936415, rs2296192, and rs9513085) that demonstrated a significant haplotype association with preeclampsia ( P = .047). Haplotype analyses of FLT1 in white women and VEGFC in both ethnic groups did not yield significant additional associations with preeclampsia.
We next compared mean serum sFlt levels between cases and control subjects, by risk alleles, given the significant associations between tagSNPs in the FLT1 and VEGFC genes in both ethnic groups. SFlt1 levels were available in 278 black women (74 cases), and 69 white women (18 cases) were evaluated ( Table 3 ). Among black and white cases, there was no significant difference in mean sFlt levels, compared with control subjects by presence of risk alleles in significant tagSNPs.
