|Home | About | Journals | Submit | Contact Us | Français|
To analyze functional polymorphisms in candidate genes (methylenetetrahydrofolate reductase (MTHFR)677C>T, MTHFR1298A>C, Factor 5 1691G>A (FVL), and angiotensinogen (AGT)-6G>A) in relation to a hypothesized placental hemorrhage pathway to preterm delivery (PTD).
We assessed maternal genotypes, pregnancy outcomes, and placental pathology among 560 white and 399 black women recruited at midtrimester into a prospective cohort study (1998-2004). Odds of dominant genotypes were calculated for PTDs with (n=56) or without (n=177) evidence of placental hemorrhage (referent=term) using race-stratified polytomous logistic regression models.
Among whites, FVL GA/AA and AGT(-6) GA/AA were both associated with hemorrhage-related PTDs (OR=4.8 [1.6, 14.2] and OR=3.8 [1.3, 10.5], respectively) but not other PTDs (ORs=1.2 and 0.9, respectively). FVL GA/AA was associated with PA (OR=5.8 [1.1, 30]) among white women. All results were null for MTHFR genotypes.
FVL and AGT variant genotypes were specifically associated with hemorrhage-related PTDs.
Preterm delivery (PTD), defined as delivery prior to 37 completed weeks' gestation, complicates more than 10% of pregnancies in the United States, and contributes to the burden of neonatal morbidity and mortality.1 Several pathways have been implicated in the etiology of PTD, including infection, stress, uterine distension, and uterine bleeding.1 Placental abruption (PA), a rare pregnancy complication characterized by premature placental detachment, is strongly associated with PTD.2 Moreover, other less severe concealed or subclinical placental hemorrhage evident only in placental pathology examinations may mark excess PTD risk3. Although it has not been established whether all manifestations of placental hemorrhage belong to a common disease pathway, we hypothesize that preterm PA cases are the “tip of an iceberg,” the most extreme manifestations of a broader hemorrhage-related pathway.
Recurrence risk for PA is very high,4, 5 potentially suggesting a role for genetic factors. Hemostatic and hemodynamic dysregulation are two vascular mechanisms that have been implicated in PA risk. Two polymorphisms in the methylenetetrahydrofolate reductase (MTHFR) gene and one polymorphism in the Factor V (F5) gene (i.e. the Leiden variant, FVL) have been linked to a thrombophilic disposition. The angiotensinogen (AGT) gene has been studied with respect to hemodynamic dysregulation. Functional polymorphisms in these three genes have all previously been investigated in relation to PA (see review and meta-analysis6). Evidence of an association between FVL and PA has been fairly consistent, while evidence of associations between MTHFR or AGT variants and PA has been less convincing. Few studies have reported on these variants in relation to PTD, and results to date have not demonstrated significant associations7-12. Other studies have attempted to link thrombophilias to vascular placental pathology findings among placentas from complicated deliveries,13-15 but these have not compared findings with term or uncomplicated deliveries. Given that some vascular function genotypes have been linked to PA, and PA is certainly linked to PTD, yet vascular function genotypes have not been convincingly linked to PTD or placental histopathology findings, we hypothesized that relevant genotypes might be associated with PTD subtypes defined by PA or hemorrhage-related placental pathology findings. We aimed to assess the associations between maternal gene polymorphisms and PTDs with clinical or pathology-based evidence of placental hemorrhage.
The Pregnancy Outcomes and Community Health (POUCH) Study enrolled women with singleton pregnancies (15-27 weeks' gestation) from 52 clinics in five Michigan communities. The POUCH Study protocol was approved by the Institutional Review Boards of Michigan State University and Michigan Department of Community Health, and all women provided informed consent. Women were eligible for the POUCH Study if they had maternal serum alpha-fetoprotein (MSAFP) screening at 15-22 weeks, no pre-existing diabetes outside of pregnancy, age ≥ 15 years, English language proficiency, and no known congenital anomalies at the time of enrollment. Women with unexplained high MSAFP (≥2 multiples of the median) were oversampled, constituting 7% of the cohort (estimated 3.5% in general obstetric population). At enrollment, women provided a blood sample, participated in an in-person interview with a trained research nurse, and filled out a questionnaire. From the interview and questionnaire, information was gathered on demographics, anthropometrics, substance use, prenatal vitamin use, and reproductive history. Women were asked to select their primary race or ethnic heritage from a list, e.g. “White or Caucasian” or “Black or African-American” (referred to as “white” and “black”).
A subcohort was selected for detailed study of biologic samples and medical records. The subcohort sample included all PTDs, all women with unexplained high MSAFP, and a stratified sample of term pregnancies with normal MSAFP; black women with term deliveries and normal MSAFP were oversampled. For the subcohort, medical records were abstracted in detail by study nurses with labor and delivery experience, placentas were examined by the study pathologist (PK Senagore), and genetic assays were conducted. The current analysis includes 996 white or black subcohort women with complete genotype and placental pathology data, and no placenta previa.
Gestational age was calculated based on last menstrual period; this estimate was replaced with an estimate from an ultrasound scan performed at <25 weeks' gestation if the two estimates differed by >14 days. PA was determined by consensus of three clinician-reviewers based on information in patient charts. The case definition required either (1) documented signs and symptoms consistent with PA (e.g. significant bleeding not attributable to dilatation, uterine pain or tenderness, fetal distress); or (2) retroplacental hematoma visualized on a prenatal ultrasound scan.
The pathologist was blinded to all clinical data including gestational age. For the gross examination, parallel slices were made through the placental disc 1 cm apart. The pathologist noted blood clots on the retromembranous, retroplacental, and disc cut surfaces, and adjacent tissue changes, i.e. dissecting hemorrhage, infarcted or compressed tissue, and red/brown tissue discoloration. Disc-impacting blood clot was defined as gross examination evidence of a retro- or intraplacental clot impacting adjacent tissue.
For the microscopic examination, nine placental tissue samples were examined: two membrane rolls, two sections of umbilical cord, and five full-thickness sections from the disc,16 without pathologist knowledge of gross examination findings. Severe histologic chorioamnionitis was determined as previously described16. Microscopic vascular-related findings that fell within five constructs adapted from a diagnostic coding tool were recorded (Kelly R. et. al., American Journal of Epidemiology, in press). For our analysis, the construct of interest was “Maternal Vascular – Disturbance of Integrity” (MV-I), which included microscopic evidence of retroplacental blood with adjacent disc disruption/compression, decidual hemorrhage in the basal plate, and decidual hemosiderin-like pigment in the membranes or basal plate. High MV-I scores (i.e. the top quintile based on the distribution among term, normal MSAFP deliveries) were previously shown to correlate with PTD risk. The MV-I construct was selected to serve not as a ‘diagnostic instrument;’ but rather, as a distributionally-defined measure that identifies a continuum of findings consistent with maternal vessel bleeding. In this study, we considered two thresholds of MV-I scores to serve as possible indicators atypical maternal vessel hemorrhage: the top decile and the top quintile.
Thus, pathology-based evidence of placental hemorrhage includes gross examination findings of a disc-impacting blood clot and/or high scores in the microscopic examination MV-I construct. When PA cases are excluded, these findings are referred to as subclinical evidence of placental hemorrhage.
DNA was prepared from peripheral blood using a Puregene (Gentra) kit. Four polymorphisms were assayed by Polymerase Chain Reaction (PCR) followed by restriction digestion with appropriate restriction enzymes. The G-6A promoter variation in AGT and the FVL variation in F5 were detected using published protocols.17, 18 The C677T polymorphism in MTHFR employed primers C677T (sense) 5′-TGA AGG AGA AGG TGT CTG CGG GA-3′, (antisense) 5′-GAC GAT GGG GCA AGT GAT TC-3′ for PCR amplification followed by digestion with HinfI to produce 100 base pair (bp) and 19 bp fragments for the C allele and 78, 22 and 19 bp fragments for the T allele. The MTHFR A1298C variant assay employed primers A1298C (sense) 5′-TCT ACC TGA AGA GCA AGT CC-3′, (antisense) 5′-CAC TTC CAG CAT CAC TCA CT-3′, followed by MboII digestion of the PCR product to yield 72, 30, 28 and 20 bp products for the A allele and 100, 30, and 20 bp products for the C allele. Minor allele frequencies and deviation from Hardy-Weinberg equilibrium were calculated by race.
Analyses incorporate weights (inverse of sampling probability) using the SURVEY procedures in SAS (Statistical Analysis Software, Cary, NC) to account for the complex sampling scheme.
As a first step, we compared distributions of maternal characteristics to evidence of placental hemorrhage using a 3-level hierarchical variable: (1) PA, (2) subclinical evidence of placental hemorrhage (i.e. disc-impacting blood clots or the top decile of MV-I scores) or (3) none. Second, with PTD as the outcome, we estimated race-specific ORs for each variant assuming dominant models, i.e. women having 1 or 2 copies of the minor allele versus with women having 0 copies, except when zero cells prohibited analysis. This was repeated with PA as the outcome.
Next, we were interested in whether maternal genotypes were specifically associated with the hypothesized placental hemorrhage-related PTD pathway. To this end, race-specific polytomous logistic regression models were developed to compare two broad PTD subtypes with term deliveries: (1) PTDs with any evidence of placental hemorrhage (i.e. PA or subclinical evidence of placental hemorrhage), and (2) PTDs without any evidence of placental hemorrhage. This analysis was repeated after changing the threshold for defining a high MV-I score from the top decile to the top quintile. Although few cases of preterm PA were available, we performed another polytomous regression analysis after separating PTDs with PA from PTDs with subclinical evidence of placental hemorrhage. MTHFR analyses were repeated after excluding women who were taking prenatal vitamins preconceptionally. All analyses were repeated after excluding women with severe histologic chorioamnionitis.
Based on a type I error rate of 0.05, the POUCH Study was originally designed to have at least 65% power to detect ORs of 2.0 in the subcohort for exposures present in 20% of the population, when considering PTD subtypes that occurred in at least 5-7% of the population. The current analysis focuses on a narrower PTD subtype, which occurred in only 2% of the population. Variation in allele frequencies and the necessity to stratify on race resulted in 80% power to detect ORs ranging from 3.0 to 6.3 depending on genotype and race (with the exception of FVL and AGT in blacks, where extreme genotype distributions precluded most analyses.).
Race-specific minor allele frequencies are shown in Table 1. The minor alleles for MTHFR C677T, MTHFR A1298C, and FVL were more common in whites than blacks. For AGT, the G allele predominated in whites while the A allele predominated in blacks. Hardy-Weinberg equilibrium was not violated for any allele in either race.
The distributions of maternal characteristics in the study sample and the prevalence of PA, subclinical evidence of placental hemorrhage, and no evidence of placental hemorrhage by these characteristics are detailed in Table 2. The weighted incidence of PTD was 10.7%. PA was clinically diagnosed in 30 women (2.1%), and subclinical evidence of placental hemorrhage was present in 9.5% (as defined based in part on a distributional cutpoint), and these were both overrepresented among PTDs (7.3% and 12.3%, P<.05). Other maternal characteristics did not differ significantly across the outcome categories. Potential confounders of the genotype-pregnancy outcome relations would need to be associated with both the genotypes and the outcomes. None of the maternal characteristics met these criteria, thus these were not used to adjust subsequent analyses.
Table 3 shows dominant genotype models for PTD. There were no significant associations between any genotype and PTD. There was an association between FVL GA/AA and PA (OR=5.8, 95% CI 1.1, 30.3) among white women (not in table).
Results of the polytomous logistic regression analysis of PTD subtypes defined broadly as those with or without any evidence of placental hemorrhage (i.e. PA and/or pathology-based evidence of placental hemorrhage) are presented in Table 4. Among white women, those with PTD and evidence of placental hemorrhage were more likely than women with term deliveries to have the FVL GA/AA genotypes (OR=4.8, 95% CI 1.6, 14.2) or the AGT(-6) GA/AA genotypes (OR=3.8, 95% CI 1.3, 10.5); the corresponding ORs for PTDs without evidence of placental hemorrhage were close null (1.2 and 0.9, respectively).
After changing the threshold of the MV-I construct from the top decile to the top quintile (which identified 1/3 of all PTDs as having evidence of hemorrhage), results were attenuated but still significant for FVL (OR=3.2, 95% CI 1.3, 8.3) but weaker for AGT (OR=1.6, 95% CI 0.8, 3.1). Results for both MTHFR variant genotypes were null in both blacks and whites. Repeating the MTHFR analyses in women who were not taking prenatal vitamins prior to conception produced similar null results.
To ensure that infection-related PTDs were not obscuring true associations between genotypes and hemorrhage-related PTDs, the models were also run after excluding women with severe histologic chorioamnionitis from the dataset, and results were similar (not shown).
Finally, we subdivided PTDs with evidence of placental hemorrhage into those with PA and those with subclinical evidence of placental hemorrhage (Table 5). FVL GA/AA was associated with both (PTD with PA OR=5.4, 95% CI 1.0, 28.2; PTD with subclinical evidence of hemorrhage OR=4.4, 95% CI 1.1, 16.9). The AGT GA/AA genotype was not significantly associated with PTD with PA (OR=2.5, 95% CI 0.5, 13.1), but was associated with PTDs with subclinical evidence of hemorrhage (OR=5.1, 95% CI 1.5, 17.2).
In this study, we investigated MTHFR C677T, MTHFR A1298C, FVL, and AGT G-6A variant genotypes in relation to PTD subtypes defined by evidence of placental hemorrhage. Although none of the genotypes were associated with PTD overall, when we brought information on delivery timing and placental hemorrhage together, we found that FVL GA/AA and AGT(-6) GA/AA genotypes were both associated with PTDs having evidence of placental hemorrhage in white women. This association was not attributable solely to clinically-detected PA cases. Notably, the observed effects were specific in that the ORs for PTDs without any evidence of placental hemorrhage were very close to the null value.
A recent meta-analysis found a significant association between FVL and PA, based on 10 studies.6 Our results corroborate a strong association between FVL and placental abruption risk, and contribute the additional finding that FVL is associated with a specific subset of PTDs that have more broadly-defined evidence of placental hemorrhage, including subclinical hemorrhage identified through placental pathology exams. Normal pregnancy alters hemostasis, shifting the balance to a relatively hypercoagulable state.19 Inherited thrombophilias may exacerbate this shift and predispose women to develop clots at the maternal-fetal interface. Although exact mechanisms are unknown, these clots may somehow facilitate rupture of decidual blood vessels, resulting in decidual hemorrhage and possibly also premature placental detachment. However, subclinical decidual hemorrhage might lead to PTD via other mechanisms.20, 21
We found that AGT(-6) GA/AA was associated with PTD with evidence of placental hemorrhage among white women. The AGT(-6) promoter polymorphism is in strong linkage disequilibrium with the AGT M235T polymorphism.22 One study conducted in Mexico identified an association between AGT M235T and preterm premature rupture of membranes12, and only two studies have reported on AGT polymorphisms and abruption risk, with conflicting results.23, 24 This variant has also been associated with preeclampsia25 and hypertension outside of pregnancy.26 PA has been consistently linked to high blood pressure,27-30 yet it is unknown whether the association between AGT genotypes and PA observed in one large study24 was attributable to hypertension. We speculate that local effects of the renin-angiotensin system in the decidual spiral arteries31 may have implications for hemorrhage-related PTD risk, possibly in the absence of systemic hypertension.
At the POUCH study's outset, MTHFR was a promising candidate gene for vascular diseases, and its variants have received a great deal of research attention in the obstetrics literature since that time. We identified no associations between MTHFR genotypes and placental abruption or hemorrhage-related PTD in black or white women. These null results add to accumulating evidence that these variants may not be strongly related to pregnancy outcomes,6-7, 9, 32-34, 36-37 although a few studies have identified positive associations with PA.10-11, 35 While there is strong biologic rationale for a role of MTHFR variants in poor pregnancy outcomes through a thrombophilia pathway, variant genotypes may only result in hyperhomocysteinemia with a thrombotic tendency among individuals with low folate intake. POUCH Study enrollment began after mandatory grain fortification in the United States, and most women reported taking prenatal vitamins, thus folate deficiency was probably rare. Direct measures of folate and homocysteine status were not available. Results did not differ meaningfully after we excluded women who took prenatal vitamins prior to conception, but we cannot rule out a stronger effect in populations with substantially lower folate intake.
The results of this study suggest additional avenues for research. Variants in other genes implicated in thrombophilias, e.g. Factor 2 (prothrombin) and plasminogen activator inhibitor-1, and other variants in renin-angiotensin system genes, might be associated with PTD with placental hemorrhage. Furthermore, given that the placenta has a fetal genotype, consideration of fetal DNA and maternal-fetal genotype interactions may shed additional light on risk of both PA and PTD.
The most important strength of this study is the objective assessment of gross and microscopic placental pathology in a large sample of preterm and term deliveries, which enabled us to investigate genotypes in relation to etiologically relevant PTD subgroups. At least three studies have attempted to link maternal or fetal thrombophilias to specific placental lesions within pregnancies complicated by PA, preeclampsia, or fetal growth restriction; however, none of these studies included a comparison group of placentas from uncomplicated pregnancies.13-15 Given the prospective design, the POUCH Study is limited by a relatively small number of PTDs with PA or other evidence of placental hemorrhage. Some identified statistically significant relations, such as that between FVL or AGT genotypes and hemorrhage-related PTDs in white women, may be accompanied by inflated ORs, as is often the case for newly discovered associations38. While we believe the results of this study are valid and biologically plausible, caution is warranted in interpreting the magnitude of the observed associations in light of the acknowledged power limitations. No adjustments were made for multiple comparions and it is likely that some identified associations would lose statistical significance if we did so; however, in our view this practice is unwarranted in the context of a priori selection of a modest number of candidate genes because it inappropriately increases the probability of type II error39.
Polymorphisms related to hemostasis and hemodynamics may be associated with PTD through pathways involving disrupted vascular integrity, ie, PA or subclinical pathology-based evidence of placental hemorrhage. Pending replication of these findings in other studies, it may be possible to identify a set of upstream markers including these maternal gene polymorphisms that discriminates women at highest risk of vascular-mediated PTD or overt PA, and to implement preventive measures to improve outcomes for both the mother and child.
The authors would like to thank the Prematurity Study Group, project director Dr. Bertha Bullen, and physician abstractors Dr. Joseph Marshall and Dr. Judith Suess. The authors would also like to acknowledge the work of Dr. Nazish Siddiqi, Dr. Judith Suess and Ms. Lynn Thelen for reviewing suspected placental abruption cases.
Sources of Financial Support: National Institute of Child Health and Human Development grant number R01 HD034543, National Institute of Nursing Research (Renewal NIH POUCH) grant number R01 HD34543, March of Dimes Foundation (Perinatal Epidemiological Research Initiative Program) Grants 20-FY98-0697 through 20-FY04-37, Thrasher Research Foundation grant 02816-7 and Centers for Disease Control and Prevention grant U01 DP000143-01.
This research was presented at the 29th annual meeting of the Society for Maternal-Fetal Medicine, San Diego, CA, January 26-31, 2009.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.