Identifying mechanisms to prolong the length of gestation, particularly in women at risk for preterm labor and delivery, will improve both maternal and fetal outcomes. Once labor has been initiated, tocolysis as currently practiced has been only poorly effective in prolonging gestations and even when effective, may only extend gestation for a few days. While tocolysis does provide sufficient time, in some cases, for the administration of antenatal steroids to enhance fetal lung development and improve neonatal outcomes, a longer prolongation of pregnancy would be required to substantially diminish other associated morbidities and the mortality associated with preterm delivery. 17-α
-hydroxyprogesterone has been shown to be effective in prolonging gestations in women who have had a previous preterm delivery, and in particular is most effective in those women who had the earliest prior delivery (8
). Nonetheless, progesterone is only able to extend pregnancy in a proportion of cases and there has not been a recognized effect of this treatment in population-based samples of women who are unselected for prior pregnancy history.
Pharmacogenetic variation in genes involved in either progesterone biosynthesis, absorption, metabolism, or function could be hypothesized as genetic candidates for identifying subpopulations who might be more susceptible to progesterone therapy, who might require larger or smaller dosages of progesterone for the therapy to be effective, or who may be untreatable by progesterone and in whom other mechanisms for the onset of preterm labor should be searched.
Humans do not demonstrate a decrease in serum progesterone levels before labor, as some mammals do. A functional progesterone withdrawal caused by changes in the progesterone receptor or expression of the receptor may be important in the initiation of labor (15
). We hypothesized that variation in the progesterone receptor might underlie some of the risk for preterm delivery and evaluated a group of polymorphic variants within PGR for these effects.
We selected SNP to capture the majority of the genetic information contained within the PGR gene. PGR is located on chromosome 11 and the gene itself is contained entirely within a large, approximately 200-kb long, haplotype block in which there is strong correlation between most of the SNP within this block. This enables a comprehensive search for genetic effects of common variants that might be present within the haplotype structure (22
). The few functional variants reported for PGR (24
) were either tested for in this report (PROGINS Alu insertion) or were in complete linkage disequilibrium with SNP that were tested [the V660L variant (rs1042838) is in complete disequilibrium with rs1042839 tested here as well as with the PROGINS insertion with 660L in cis with the PROGINS Alu insertion]. Of the four SNP showing p
< 0.05 in either fetal or maternal testing done here, two, showing significance in the mother (rs653752 and rs503362), are in complete disequilibrium with the V660L and the PROGINS insertion, suggesting that these variants might be providing the etiologic mechanism underlying the observation seen here. Because 660L and PROGINS Alu insertion allele is less responsive to progestin than the 660V/no insertion allele (24
), the association described may have a basis in this functional difference. We also identified strong evidence of linkage disequilibrium and presumed association using the singleton fetus/infant as a risk case with the strongest pairwise signals identified in SNPs approximately 15 and 25 kb 5′ of the PGR promoter. These other two SNP with p
< 0.05 (rs1942836 and rs4754732) are in incomplete disequilibrium with the PROGINS variants so their role is less certain. However, these two SNP are located in possible regulatory regions suggesting that a second mechanism could be acting as well, and there may be different mechanisms in the mother than in the fetus.
Our results were less statistically significant when data from twin pregnancies were included. Because twin pregnancies are more likely to deliver before 37 wk gestation, their inclusion in the analysis creates challenges, including at what gestational age they should be considered physiologically premature. At the same time, dose response effects modulated by the fetus might be more striking in twins (with twice the dose of gene products of the twins) and these effects might be evaluated in future analyses. Our use of twins was considered exploratory in this project and we anticipate that additional analytic tools will be required to properly and comprehensively make use of twins in genetic analyses of preterm labor.
We also undertook a comprehensive sequencing study of the PGR gene in mothers and focused on coding exons, splice sites, and a few highly conserved regions that might function as regulatory elements. No high-risk variants were identified in this search, although several rare variants, of yet unknown etiology, create opportunities for investigation. We assumed that if the effect of genetic variants was fetally mediated, these would be present at one half their expected frequency in the mothers of preterm infants and thus chose for reasons of efficiency to sequence only the mothers.
The PGR gene is expressed as two isoforms that are generated via
alternate promoter elements 5′ of the coding sequence of the gene (33
). PGR is regulated in utero
in the fetal membranes with isoform expression for the two common forms progesterone receptor isoform A (PR-A) and progesterone receptor isoform B (PR-B) having reciprocally higher expression before and after labor (35
). The PR-A form seems to behave as a repressor of PR-B in the amnion as well (36
). Similarly, in the myometrium, PR-A may initiate or modulate progesterone withdrawal to stimulate the onset of labor (37
). Additional investigations can now focus on identifying polymorphic variants within specific conserved elements to determine whether they may play a role in altered expression of the PGR receptor and to carry out additional investigations on the functional elements at V660L and PROGINS for the associations described.
Weaknesses of this study include the confounding effects of using a population-based collection of preterm labor cases where there is likely underlying causal heterogeneity, using the conservative Bonferroni method to correct for multiple comparisons and only indirect functional correlates with the associated genotype. Since these were registry-based samples, there was little data available on maternal indications for preterm delivery so that we were unable to stratify on this variable. The heterogeneity, however, should only predispose to false-negative results, not false positives and the characterization of functional correlates will await further study. Using the conservative Bonferroni correction, p values of formal significance were not seen, but as an initial exploratory investigation, several regions in the PGR gene are now identified for further investigation using larger numbers of samples and more defined phenotypes. In addition, Bonferroni is highly conservative and does not consider the linkage disequilibrium relationships across the PGR locus, which could serve to modulate the effects seen. Thus, the haplotype association may well be indicative of genuine associations.
Strengths of this study are the large sample size used, the focus on the fetus and mother as risk cases, and the extensive genotyping of the PGR gene. We also incorporated a DNA sequencing approach, which allows for detection of rare variants if the common disease/common variant hypothesis is not satisfied, as must be the case when an association-based approach is undertaken. Because many genetic disorders show substantial allelic heterogeneity, using both association and sequencing provides a comprehensive survey for variants that might be disease contributing (38
). Genome-wide approaches using linkage and disequilibrium will also allow for finding those variants not suspected by our current understandings of the biology of parturition (39
). The genetic effects identified here hold promise for use in future clinical studies to determine whether mothers or their fetuses with particular PGR genetic backgrounds might be more effectively treated with progesterone. Alternatively, there might be a subpopulation genetically resistant to progesterone and in whom alternative therapies should be tried first. A better characterization of the biologic underpinnings of these observations will provide opportunities to generate new therapeutic and preventative options for preterm labor.