Background
The increasing prevalence of cardio-metabolic diseases leading to premature death is a raising concern [
1]. Although major risk factors are lifestyle factors after birth, epidemiological studies have associated low birth weight (LBW) and/or fetal growth restriction (FGR) with hypertension, cardiovascular diseases, and type 2 diabetes, indicating the existence of prenatal risk factors [
2‐
4]. However, the underlying mechanisms explaining the relationship between LBW/FGR and diseases in adulthood are not fully understood. It is suggested that maternal undernutrition and/or poor placental function could alter long-lasting body function and physiology, predisposing individuals to cardio-metabolic diseases [
5,
6]. To analyze the effect of such intrauterine environment on fetal growth, the maternal genotype has been used as a proxy for the intrauterine environment [
7,
8]. Maternal genetic factors may influence fetal growth directly through the alleles inherited by the fetus, or indirectly through the intrauterine environment, and this idea has been developed into the Mendelian Randomization (MR) approach in the framework of causal inference [
7‐
10].
Recent MR studies have reported that maternal systolic blood pressure (SBP)-increasing polygenic score (PGS) causes lower birth weight through elevated maternal SBP, independently of direct fetal genetic effects [
8,
10]. However, the genetic estimate of the association between maternal SBP and birth weight was in the opposite direction to the observational estimate [
8]. In addition, hypertension usually develops in late life. Therefore, it may not be appropriate to apply to pregnant women the estimates of lifelong BP risks in the MR approach. Thus, it remains unclear why higher maternal SBP-increasing PGS results in lower birth weight.
FGR pathophysiology and etiology are complex, but most cases are thought to arise from placental dysfunction [
4,
11]. Generally, placental growth precedes fetal growth, with the placental growth rate decreasing from about 30 weeks, and the fetus/placenta growth ratio gradually increasing until 37–38 weeks [
12]. Nearly 80% of FGR are late-onset (≥ 32 weeks) [
13]. In addition, it has been highlighted that placental weight is an important determinant of fetal growth in the third trimester [
14], and birth and placental weight highly correlate [
14,
15]. The placenta is a vascular organ, and recent GWAS uncovered numerous BP-associated genes involved in regulating the development and function of the vascular system [
16‐
18]. Therefore, we hypothesized that BP-increasing maternal SNPs influence placental growth, which in turn affects birth weight.
Since the mediating role of the placenta was never been examined, we investigated whether maternal SBP-increasing PGS is inversely associated with placental weight and whether the placental weight mediates the effect of maternal SBP PGS on birth weight. Furthermore, by constructing a genetic BP-increasing risk score composed exclusively of vasculature-related SNPs, we aimed to clarify that the role of placental weight in the effects of vasculature-related BP SNPs on birth weight. We also aimed to confirm by haplotype analysis that maternal transmitted and non-transmitted BP-increasing alleles, but not paternal transmitted alleles, are inversely associated with birth and placental weight. Lowering of birth weight can be caused by deceleration of fetal growth velocity during pregnancy. Motivated to understand the indirect effect of maternal SBP PGS on fetal growth velocity, reflecting the fact that placental growth precedes fetal growth, we aimed to determine that maternal SBP PGS significantly decreases fetal growth velocity at the very end of pregnancy. Thus, our study uncovered that maternal genetic risk of hypertension inhibits fetal growth through the intrauterine environment established by the placenta.
Discussion
This is the first study to show that placental weight mediates the effect of maternal SBP-increasing PGS on lowering birth weight. While we found a notable inverse association between maternal genetic score and placental weight, observational BP traits were not associated with BP genetic score. In addition, placental weight well correlated with birth weight, whereas no correlation was observed between observational BP and birth weight. PGS constituted of BP SNPs only related to the vascular system also had a reducing effect on birth weight, and this effect was mediated by the placental weight. Additionally, we put in evidence that both maternal transmitted and non-transmitted alleles of BP-increasing SNPs, but not paternal transmitted alleles, affect birth weight. The association between maternal SBP PGS and deceleration of fetal growth velocity only appeared later in gestation, consistent with a placental mediation of the effect of maternal hypertension genetic risk on birth weight.
Previous MR studies of the European ancestry population have consistently reported an inverse association between maternal SBP-increasing PGS and birth weight [
7,
8,
10]. In the present study based on Japanese pregnant women, we replicated this association with linear regression analyses, although SNPs constituting PGS for the Japanese population differed from those for the European population.
Using MR analysis, Tyrrell et al. estimated that every 10 mmHg increase of the maternal SBP is genetically associated with a 208-g decrease in birth weight [
8]. However, they observed that a 10-mmHg increase in measured maternal SBP was associated with a 24-g increase in birth weight [
8]. That is, the direction of the observed association was opposite to the estimated genetic association. This was inconsistent with the assumption that SBP traits mediate the effect of maternal SBP genetic score on birth weight, but it was thought to be attributed to confounding, such as BMI. Since then, this issue has not been fully explained. Our results indicate that it is unlikely that maternal BP traits mediate the effect of maternal BP PGS on birth weight, even when pre-pregnancy BMI is included as a confounder in the analysis. In contrast, the present study found that maternal SBP increasing genetic score was associated with reducing placental weight, and placental weight correlated with birth weight. A mediation analysis showed that more than 70% of the effect of maternal SBP increasing PGS on birth weight was mediated by placental weight. The results of the mediation analysis were not altered by the exclusion of SNPs associated with BMI, HbA1c, and T2D among the SBP-associated SNPs. Furthermore, the result of our sensitivity analysis for sequential ignorability [
43,
44] also showed that even if there were unmeasured confounders, they would not affect the conclusion that the placenta mediates the effect of SBP PGS on birth weight.
Genes associated with hypertension have been identified by recent GWAS [
16‐
18], and beyond genetic variants that work in the kidney and adrenal glands, many SNPs have been shown to be related to vascular function, vascular development, angiogenesis, and remodeling. This prompted us to subdivide BP SNPs into either vasculature-related or unlikely-related groups and examine the association of their genetic scores with birth weight. We observed that the genetic score of SNPs related to the vascular system presented a higher association with birth weight, and this was mostly mediated by placental weight.
To analyze the maternal genetic score as a proxy for the intrauterine environment, Beaumont, Warrington, and their colleague developed a method of analysis that distinguishes it from the effects of genetic factors shared by the fetus [
7,
10]. They showed that the effect of the SBP-increasing alleles on lowering birth weight originated from maternal alleles. In the present study, we used a method [
36] to calculate genetic scores separately for maternal transmitted, maternal non-transmitted, and paternal transmitted alleles, and confirmed that the effects of BP-increasing alleles on birth weight and placental weight were only observed for maternal alleles. The results are in agreement with previous studies [
7,
10].
Fetal growth restriction (FGR) is characterized as a growth deceleration from a certain point in the gestation period, resulting in fetus pathologically small for the gestation period (SGA), which is different from constitutionally SGA [
4,
46]. As most FGR are thought to arise from placental dysfunction, which matures in the 3rd trimester and slows down its growth after around 30 weeks, it appears reasonable that nearly 80% of FGR are late-onset (≥ 32 weeks) [
13]. When and how the fetal growth velocity changes are important information. We previously calculated the EFW increase per week (differential value) every week, which is called fetal growth velocity, for individual fetuses and analyzed the trajectories of the velocity. We have shown that the heterogeneity of growth velocity trajectories is most pronounced after 30 weeks of gestation [
20]. Individual trajectories analysis highlighted the continuity of the degree of growth restriction, and ≥ 10% of the fetuses showed a third-trimester growth deceleration. Considering the time sequence of events, maternal PGS should be associated with fetal growth velocity in late gestation where placental maturation is proceeding. As expected, the inverse association between maternal SBP-increasing PGS and fetal growth velocity became progressively apparent towards 36 weeks. This result, although indirect, further supports a placental mediation of the effects of maternal BP genetic score on birth weight.
The NICHD Fetal Growth Study, which explored the association between maternal SBP (or DBP)-increasing PGS and EFW for each trimester, has been recently reported [
47]. A unique feature of the NICHD study is that ethnically different populations (Hispanic, non-Hispanic White, non-Hispanic Black, and Asian) were included in the analysis. The authors referred to the GWAS summary statistics of European ancestry to construct PGS and separately applied it to all distinct target populations. However, it is well known that PGS analyses using the summary statics of ethnically different population from the target population are not reliable [
27,
48]. In addition, since the frequency of ultrasound measurements was not very high, the estimation of each individual gestational week-specific EFW might have not been accurate. In our analysis, the study target population was limited to Japanese, and PGS analysis was performed using summary statistics of Japanese GWAS to obtain reliable results. In addition, we were able to collect a much larger number of ultrasound measurements for every fetus so that we could obtain weekly EFW values and their differentiated value, fetal growth velocity [
20]. We focused our analysis on the association between the gestational week-specific fetal growth velocity and maternal SBP PGS, and we found that this association only became apparent late in gestation.
Up to now, maternal genetic scores associated with SBP, DBP, BMI, and T2D or birth weight-associated SNPs were used as a proxy for the intrauterine environment to validate Developmental Origin of Health and Disease (DOHaD) hypothesis [
7‐
10]. Recent MR studies have concluded that the intrauterine environment is unlikely to be a major determinant of adverse cardiometabolic outcomes in offspring [
10]. However, our study reveals a significant association between placental weight and maternal PGS composed of partial BP SNPs related to the vasculature. This result suggests that it is possible to design a maternal genetic score representing the intrauterine environment much better. A large placental weight GWAS using mother and child genome will be needed to investigate the association between the maternal genetic score related to placental weight and the offspring’s birth weight and cardiometabolic traits in adulthood. Alternatively, since placental function may not necessarily correlate with placental weight, we need to devise a way to evaluate the genetic score that more directly reflects the intrauterine environment. It will also be interesting to see what the impact on the cardiometabolic outcome of the offspring will be in MR approach if maternal BP SNPs are limited to the vasculature and/or placenta-related SNPs. In any case, we think more analyses need to be performed before intrauterine programming is ruled out.
The main strength of our study was using the cohort of mother-offspring pair cohort that contains unique phenotype data for time-series fetal growth velocity and placental weight as well as genotype. In addition, birth weight and placental weight were ideally adjusted for gestational age, fetal sex, and maternal parity to eliminate well-known confounding factors. Pre-pregnancy BMI was also considered as a confounding factor. Using causal mediation analysis, we found for the first time that the effects of maternal SBP PGS on birth weight are mediated by placental weight. However, some limitations should be noted. First, the sample size of the cohort with available genotype was small, around 100, which is the minimum target sample size for PGS analysis [
27]. The association between SBP-PGS and placenta was much stronger than the association between typical PGS and the corresponding traits; the variance of placenta trait explained by the PGS, represented by adjusted
R2, exceeded 0.2; therefore, we were able to perform statistically reliable association analysis with this sample size. Nevertheless, the results of this study will need to be replicated in another cohort possessing complete data of fetal growth velocity and placental weight. Second, we classified PGSs based on literature and open-source data on gene activities and chromatin status in vascular tissues for the genes affected by SNPs. Although these methods were commonly used in previous studies [
16‐
18], some information is not yet complete and will improve as more data accumulate.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.