Abstract
Background: Circulating relaxin levels have repeatedly been associated with preterm birth (PTB). Our aim was to investigate if mothers carrying promoter single nucleotide polymorphisms (SNPs) in one of the three relaxin genes (RLN1, RLN2, RLN3) are predisposed for PTB. Patients and Methods: Maternal DNA from 80 preterm cases (40 very preterm births (24-34 weeks) and 40 moderate preterm (34-36 weeks) and 40 controls (term delivery)) nested in the Danish National Birth Cohort were examined for nine SNPs. Results: Maternal homozygosity of the rarer allele in the relaxin 2 gene (RLN2, rs10115467 and rs4742076) had an increased risk of moderate PTB (odds ratio 4.1 [95% CI 1.4-12] and 8.8 [95% CI 1.03-75] respectively). Only rs10115467 remained significant after correction for multiple testing. Conclusion: Women homozygous for prevalent SNPs in the RLN2 gene may have a genetic susceptibility for PTB.
Preterm birth (PTB) is associated with much morbidity and mortality. Identification of women at risk of spontaneous preterm birth may be an important step toward our understanding of PTB pathophysiology and for possible clinical interventions. To provide a time window for potential interventions, the identification of women at risk is preferable early in pregnancy. To date, early risk assessment has met with little success.
A possibility of genetic predisposition to PTB has repeatedly been proposed. Twin studies assess the heritability of preterm birth to around 25 % (1, 2) and additional studies suggest that genetic risk factors play an important role in the pathogenesis of PTB (3-5). A number of case-control studies have identified potential candidate genes associated with PTB over the last years (6).
Relaxin is a hormone produced in the corpus luteum during pregnancy. Its effect on the cervix is well described in animals, where it is involved in collagen remodeling, pubic symphysis and mammary gland development during pregnancy (7). Moreover, relaxin inhibits uterine contractions in animals (7). Relaxin's role in human parturition is still largely unknown, but studies have shown that it has no effect on the myometrium, and that the profile of circulating relaxin is very different in humans compared to animals (7). In relevance to PTB, high levels of circulating relaxin have been shown to be associated to PTB (8-10). In the human cervix relaxin receptors have been found in fibroblasts, and relaxin has been observed to stimulate the activity of metalloproteinases (11-13). Human relaxin induce a weakening of the fetal membranes in vitro (14) and porcine relaxin may induce cervical maturation (15). This effect could not be reproduced using human relaxin (16). Previously we found circulating relaxin to be associated with cervical maturation assessed by digital examination and with preterm prelabor rupture of membranes in women with threatening PTB (17). To this point an association between relaxin levels and short cervical length measured by ultrasound has not been established, even though both are individual predictors of PTB (8).
We are not aware of any data regarding SNPs in the human relaxin genes. The purpose of the present study is to evaluate whether SNPs in the promoter region of the relaxin genes are associated with PTB. We therefore examined allele, genotype and haplotype frequency differences in preterm and term deliveries for nine SNPs in three relaxin genes: RLN2 (circulating relaxin), RLN1 and RLN3.
Patients and Methods
This case-control study was nested within the Danish National Birth Cohort (DNBC), which recruited over 100,000 women early in pregnancy between 1997 and 2002. During pregnancy, women completed two interviews and provided two blood samples. Methods for the DNBC study are described elsewhere (18).
Fourty women with deliveries before 34 weeks gestation and 40 women with deliveries between 34 and 37 weeks gestation were randomly drawn from the DNBC cohort along with 40 women with deliveries at/after 37 weeks gestation serving as term controls. Due to the small sample size, data analyses were primarily made in the whole dataset (40 very preterm, 40 moderate preterm and 40 controls); subsequently analyses were confirmed in a smaller and restricted dataset. For the restricted dataset we excluded subjects if they: i) delivered an infant with a birth defect; ii) had a multiple gestation, or if iii) delivery was indicated. Birth defects were in majority defects associated preterm delivery (persistent fetal circulation (n=2), septal heart defects (n=2), non-descended testis (n=1), hip dysplasia (n=2), hydrocephalus (n=3), polycystic kidney (n=1)). The smaller dataset thereby consisted of 27 very preterm, 30 moderate preterm and 40 controls.
Gestational age was estimated from self-reported last menstrual period, but corrected with an early ultrasound estimate if the subject used contraceptive in the four months prior to conception, had irregular periods, or had an abnormal last menstrual period. DNBC-supervised personnel administered detailed telephone interviews to each woman at 12 and 30 weeks (available at http://www.ssi.dk/sw9314.asp). Additionally, EDTA-treated whole blood was collected at the first prenatal appointment. Samples were subsequently mailed to the Statens Serum Institute in Copenhagen, and upon arrival a part of the blood was transferred to filter paper and the surplus separated into plasma and buffy-coat. We have previously demonstrated that 79 percent of all blood samples arrived for processing and storage within one day of collection (19). All samples were stored at −20°C after reception.
Genomic DNA from all dried blood spots was extracted from one 3.2 mm disk in a 200 μl volume using the Extract-N-Amp Blood PCR Kit (Extract-N-Amp, Sigma-Aldrich, St. Louis, MO, USA). The gDNA was then quantified using the Quant-iT™ PicoGreen® dsDNA Reagent (PicoGreen, Molecular Probes, Invitrogen, Carlsbad, CA, USA). Both extraction and quantification was carried out according to the manufacturers' protocols. The gDNA was amplified using the GenomePlex Whole Genome Amplification Kit (Sigma-Aldrich). The DNA storage and extraction method has previously been validated (20).
Genotyping of the nine RLN SNPs (RLN1- rs3758240, rs1322220, rs10481591, rs7048887, rs1575279; RLN2- rs3758239, rs4742076, rs10115467; RLN3- rs7248735) was conducted using a custom Illumina GoldenGate genotyping assay consisting of 1,152 SNPs scattered throughout selected regions of 140 potential preterm birth or cerebral palsy candidate genes. The relaxin gene SNPs were chosen based on the hypothesis that SNPs in the regulatory region of the genes induce abnormal expression of relaxin proteins and thereby induce PTB. The biobank samples and the Illumina assay (Illumina, San Diego, Ca, USA) are described and evaluated in more detail elsewhere (20).
The analysis of the data from this case-control study is as follows: At first we analyzed the frequency of the SNPs. One SNP (the only SNP in the RLN3 gene) had an allele frequency hovering at just 5% and was excluded from further analyses. Secondly, we tested for Hardy-Weinberg equilibrium in controls at 5 % alpha level, and all analyses found Hardy-Weinberg equilibrium. Subsequently the data were tested for linkage disequilibrium by calculating R2. Three SNPs in the RLN1 gene (rs1322220, rs10481591 and rs7048887) were in high linkage disequilibrium (R2>0.9). As this indicates redundancy, two SNPs were excluded and only rs1322220 remained for individual SNP association analyzes. Except for the haplotype analyzes, the subsequent analyses will therefore be completed for three SNPs in the RLN1 gene (rs1322220, rs3758240, rs1575279) and three SNPs in the RLN2 gene (rs3758239, rs4742076, rs10115467). Case-control distributions on discrete variables were characterized by proportions and odds ratios with 95% confidence intervals (95% CI). P-values in association tests were adjusted for multiple testing by permutation testing (step-down min P) (21). Dominant, recessive and additive (dose-response effect of the mutant allele) modes of effect were considered and the best fitting model was chosen. Haplotypes were estimated and tested for association using the likelihood-based methods of Lin et al. (22).
Results
Maternal characteristics are described in Table I, and none of the covariates were distributed unevenly among strata. Table II describes the observed genotypes for the selected loci, and their associations with PTB are described in Table III.
SNP associations with PTB. Mothers who are homozygous of the rare allele (RLN2 rs10115467) were at a significantly increased risk of moderate PTB (34-36 weeks gestation) (odds ratio 4.1 [95% CI 1.4-12], permutation p-value=0.03), Table III. The risk of very PTB (<34 weeks) was not significantly increased (OR 1.4 [95% CI 0.4-4.4], permutation p-value=0.84). Mothers homozygous for the rarer allele of a different SNP in the RLN2 gene (rs4742076) were also at a significantly increased risk of moderate PTB (34-36 weeks gestation) (odds ratio 8.8 [95% CI 1.03-75]). However, when considering multiple testing, it did not remain significant (permutation p-value=0.10). The odds ratio of very PTB (<34 weeks) for the same allele was not significant, but indicated a greater risk (OR 4.3 [95% CI 0.5-41], permuted p- value=0.62).
The RLN2 (rs3758239) and the RLN1 SNPs (rs1322220, rs3758240, rs1575279) were not statistically significantly associated with PTB.
Haplotype associations with PTB. The associations between PTB and unphased haplotypes were established (Table IV). The wildtype haplotype was used as the referent category. Only one haplotype (the combination of SNPs rs3758239 and rs10115467 in RLN2) was associated with a significantly increased risk of moderate PTB (OR 2.8 [95% CI 1.01-7.9]).
In the restricted dataset a narrower phenotype (spontaneous PTB, no multiples, no malformations) was used. Odds ratios for individual SNP analyses as well as haplotype analyses were similar on the restricted dataset (27 very preterm, 30 moderate preterm and 40 controls), but returned wider confidence intervals due to the smaller sample size.
Characteristics of women with very preterm and moderate preterm deliveries compared to controls: a nested case-control study from the Danish National Birth Cohort.
Observed genotype for specific SNP loci in relaxin genes among women with preterm (n=40+40=80), very preterm (n=40) and moderate preterm (n=40) births. ‘AA’, ‘AB’ and ‘BB’ represent the three possible genotypes at each locus.
Discussion
Single nucleotide polymorphisms in the human relaxin genes have not previously been examined or linked with PTB. We show that women who are homozygous for common SNPs in the RLN2 gene may have a genetic susceptibility for PTB. The finding of genetic susceptibility to PTB in the RLN2 gene is very interesting. Firstly, it provides support for the involvement of relaxin in the etiology of PTB. Secondly, it adds to the body of evidence that genetics are important in birth timing (6). Further examination of these SNPs in the RLN2 gene may demonstrate if these markers are of predictive importance in PTB. The high odds ratios of PTB for both RLN2 SNPs could indicate this. Genetic maternal markers are ideal as predictors of pregnancies at risk for PTB, as identification of women at risk becomes possible even prior to pregnancy.
The function of relaxin in human pregnancy is largely unknown. In animals, relaxin is known to promote of growth and dilation of the cervix, growth and quiescence of the uterus, growth and development of the mammary gland, and regulation of cardiovascular function (7). Whereas relaxin levels in animals rise throughout pregnancy and peak prior to the initiation of labor (7), the human profile is different. In an uncomplicated human pregnancy, the relaxin levels peak in weeks 8-12, and then decline and stabilize at lower concentrations (10, 23). High serum relaxin levels throughout second and third semesters are associated with PTB (8-10). An increased risk of PTB has also been shown in pregnancies after assisted fertilization (24) and twin pregnacies (25), two types of pregnancies that are known to have elevated relaxin levels (10, 26) and an increased risk of PTB.
Little is known about the regulation of relaxin (27). A study on rat cultured granulosa cells showed that prolactin can induce expression of relaxin, with estrogen acting as an enhancer (28). Other studies indicated that luteinizing hormone (29) and progesterone E2 (30) can stimulate the secretion of relaxin, and that basic fibroblast growth factor may work as an inhibitor. Additionally a human study in showed that the human chorion gonadotropin hormone can stimulate corpus luteum to express relaxin (31). Finally, both progesterone and glucocorticoids can regulate relaxin 2 expression in human choriocarcinoma cells (32). The regulation of the pregnancy hormones is very delicate, and one dysregulated hormone is likely to be counterbalanced by others. This may tend to minimize the effect of expression-affecting SNPs.
Single SNP association analysis with preterm birth.
The associations between unphased haplotypes for Relaxin-2 (RLN2 SNPs rs3758239, rs4742076, rs10115467), Relaxin-1 (RLN1 rs3758240 rs1322220 rs10481591 rs7048887 rs1575279) and preterm birth.
We hypothesize that SNPs in the relaxin promoters might cause the relaxin genes to be dysregulated, thereby affecting the relaxin levels in pregnancy. Changes in the relaxin profile is known causes a predisposition to PTB (10). We hypothesize that as in animals this effect of relaxin is caused by an inhibition of collagen expression, the up-regulation of matrix metalloproteinases and through fibroblast activation. Relaxin hereby causes a softening of the birth canal late in pregnancy. This predisposition by altered relaxin expression is only resulting in PTB if additional factors are present such as: biomechanical factors (a weak cervix due to low hydroxyproline content, a previous conisation, a large physical strain due to twin pregnancy, a strenuous job or obesity), biochemical factors (diethylstilbestrol, smoking, progesterone/estrogen ratio), exogenous factors (infection, trauma) as well as fetal factors (intrauterine growth retardation, birth defects, genetic predisposition). This increases the need for large sample sizes but also a multifactorial approach for future studies.
Our study is drawn from a very homogeneous population, and it is not very likely that these results are due to ethnic differences in PTB and SNP frequencies. Additionally, the nested case-control design limits the risk of bias. Our study is small, and warrant replication. Studies with measurements of circulating relaxin levels to validate that hypothesized regulatory effect of these SNPs would be highly valuable. In conclusion, we have found two SNPs in the regulatory region of the RLN2 gene to be statistically significantly associated to an increased risk of PTB. This provides preliminary evidence of the involvement of human relaxin in the etiology of preterm birth.
- Received December 31, 2008.
- Accepted August 31, 2009.
- Copyright © 2009 The Author(s). Published by the International Institute of Anticancer Research.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.