An optimal periconception maternal folate status for embryonic size: the Rotterdam Predict study

Authors


Abstract

Objective

To investigate the association between periconception maternal folate status and embryonic size.

Design

Prospective periconception cohort study.

Setting

Erasmus University Medical Centre, Rotterdam, the Netherlands.

Population

Seventy-seven singleton pregnancies recruited in 2009 and 2010.

Methods

We recruited women before 8 weeks of gestation and performed weekly three-dimensional ultrasound scans from enrolment up to 13 weeks of gestation. As a measure of embryonic growth, crown–rump length (CRL) measurements were performed using V-Scope software in the BARCO I-Space. Maternal blood was collected to determine first-trimester long-term red blood cell (RBC) folate status. Non-malformed live births were included in the analysis. We calculated quartiles of RBC folate, square root-transformed CRL data and performed multivariable linear mixed model analyses.

Main outcome measures

Serial first-trimester CRL measurements.

Results

In total, 484 ultrasound scans were performed in 77 women, in 440 (90.7%) of which CRLs could be measured. RBC folate in the third quartile (1513–1812 nmol/l) was significantly associated with an increased CRL compared with the first two quartiles (814–1512 nmol/l) and the upper quartile (1813–2936 nmol/l; Poverall = 0.03; adjusted for gestational age, smoking, body mass index and fetal sex). Compared with the third quartile, embryos in the upper quartile were 24.2% smaller at 6+0 weeks [4.1 mm (95% confidence interval 3.5, 4.7) versus 5.4 mm (95% confidence interval 4.8, 6.1)] and 7.6% smaller at 12+0 weeks [55.1 mm (95% confidence interval 52.9, 57.3) versus 59.6 mm (95% confidence interval 57.4, 62.0)] of gestation.

Conclusions

This study suggests that a very high maternal periconception folate status is associated with reduced embryonic size. Whether these effects are beneficial or harmful requires further investigation.

Introduction

Fetal growth is associated with health and disease risks in later life.[1] Fetal growth is influenced by a multitude of genetic and environmental factors, including maternal folate status.[2] Folate is an important substrate of one-carbon metabolism, in which one-carbon groups are provided for essential cellular processes, such as the synthesis of proteins, lipids, DNA and RNA, and the methylation of chromatin.[3] As pregnancy is a period of rapid growth and numerous cell divisions, folate requirements are increased throughout pregnancy. In addition, periconceptional use of folic acid (FA) supplements has been shown to prevent neural tube defects (NTDs), which has led the World Health Organization to recommend FA supplement use from the periconception period up to 12 weeks of pregnancy.[4, 5] As a result of the high stability and bioavailability of the synthetic FA form, maternal folate status is strongly determined by FA supplement use. However, dietary folate intake, metabolism, the use of medication, certain lifestyles and health conditions, and genetic variations in folate genes, such as the gene encoding the enzyme methylenetetrahydrofolate reductase, also affect folate status.[6, 7] Therefore, folate biomarkers provide a more precise estimation of folate status. Although serum and plasma folate levels are subject to daily fluctuations in FA and dietary folate intake and represent short-term folate status, red blood cell (RBC) folate represents long-term folate status, reflecting the previous 2–4 months, as RBCs only accumulate folate during erythropoiesis and have a life span of approximately 120 days.[8] Thus, in early pregnancy, maternal RBC folate reflects the folate status in the periconception period.

First-trimester maternal RBC folate has been positively associated with weight and head circumference in newborns.[9-12] Although FA supplements are recommended to be used predominantly in the periconception period, and all major organ systems are developed within the first 10 weeks of gestation, to date no studies have been performed on associations between maternal RBC folate status and embryonic growth trajectories in early pregnancy.

Therefore, in this study, we investigated whether maternal RBC folate levels are positively associated with first-trimester embryonic growth, as determined by the crown–rump length (CRL).

Methods

Data for this study were collected in the Rotterdam Predict study, a prospective periconception cohort study conducted at the Department of Obstetrics and Gynaecology, Erasmus University Medical Centre, Rotterdam, the Netherlands. This study has been approved by the Central Committee on Research in The Hague and the local Medical Ethical and Institutional Review Board of the Erasmus University Medical Centre. At enrolment, all participants signed a written informed consent form before participation.

All women of at least 18 years of age with ongoing intrauterine singleton pregnancies of 6–8 weeks of gestation were eligible for participation, and were recruited in 2009 and 2010. In a subgroup, first-trimester maternal RBC folate was determined at enrolment. The majority of participating women were recruited from the outpatient clinic of the Department of Obstetrics and Gynaecology at the Erasmus University Medical Centre, and a smaller group (23%) was recruited from outside the hospital. For the current study, we only included those pregnancies in which first-trimester maternal RBC folate was determined.

Women received weekly transvaginal three-dimensional (3D) ultrasound scans from enrolment up to the 13th week of pregnancy. Ultrasound scans were performed with a 6–12-MHz transvaginal probe using GE Voluson E8 equipment and 4D View software (General Electrics Medical Systems, Zipf, Austria). The scanning time per visit was as short as possible (less than 20 minutes) and the thermal and mechanical indices were kept below one, in line with the international recommendations for safe scanning.[13-15] The 3D datasets obtained were stored and transformed to Cartesian (rectangular) volumes afterwards, to be transferred to the BARCO I-Space (Barco N.V., Kortrijk, Belgium) at the Department of Bioinformatics, Erasmus University Medical Centre, Rotterdam, the Netherlands. This is a four-walled CAVE™-like (Cave Automatic Virtual Environment) virtual reality system, allowing depth perception and interaction with the projected images.[16] CRL measurements were performed offline using the I-Space and V-Scope software,[17] and by placing virtual callipers at the outer side of the crown and rump in the mid-sagittal plane. CRL measurements performed in the I-Space show good agreement with 2D measurements and good inter- and intraobserver agreement.[18] All CRL measurements were performed three times by the same researcher, and the mean of these three measurements was used in the analyses.

At enrolment, participants completed a self-administered general questionnaire covering details on maternal age, anthropometrics, ethnicity, education, obstetric history and periconception exposures.

In addition, a venous blood sample was collected to determine first-trimester maternal RBC folate levels. Blood was collected in an 8.5-ml Vacutainer ethylenediaminetetraacetic acid (EDTA) tube (BD Diagnostics, Plymouth, Cornwall, UK). Directly after blood sampling, the haemolysate was prepared by diluting 0.1 ml of full blood in 0.9 ml of freshly prepared 1.0% ascorbic acid. Subsequently, the haematocrit of the remaining EDTA full blood was determined on a Sysmex XE-2100 Haematology Analyser (Sysmex, Europe GmbH, Norderstedt, Germany). In serum, folate was measured using an electrochemiluminescence immunoassay (Modular E170, Roche GmbH, Mannheim, Germany). The haemolysate was centrifuged at 1000 g for 5 min at 18°C, just before the folate measurement. The haemolysate folate concentration was recalculated into RBC folate concentration using the following formula: (nm haemolysate folate × 10/haematocrit) − [nm serum folate × (1 − haematocrit)/haematocrit] = nm RBC folate.

Data on the first day of the last menstrual period (LMP) and on the regularity and duration of the menstrual cycle were obtained in a personal interview by the researcher performing the ultrasound at the first visit. We calculated the gestational age from the LMP in spontaneously conceived pregnancies, from the date of oocyte pickup plus 14 days in pregnancies conceived through in vitro fertilisation with or without intracytoplasmic sperm injection (IVF/ICSI), from the LMP or insemination date plus 14 days in pregnancies conceived through intra-uterine insemination (IUI), and from the day of embryo transfer plus 17 or 18 days in pregnancies originating from the transfer of cryopreserved embryos, depending on the number of days between oocyte pickup and cryopreservation of the embryo. When the menstrual cycle was regular, but more than 3 days different from 28 (28 ± >3 days), we adjusted the gestational age for the duration of the menstrual cycle.

From the total number of 102 pregnancies in which first-trimester maternal RBC folate was determined, we excluded pregnancies conceived by oocyte donation (n = 2), pregnancies ending in a miscarriage (n = 12), pregnancies in which the first day of the LMP was missing or the observed CRL differed by more than 6 days from the expected CRL according to the Robinson curve[19] (n = 6) and pregnancies that ended in a major malformation with (n = 2) or without (n = 2) subsequent termination. Of the remaining women, only one reported no FA supplement use and was therefore also excluded, resulting in a total of 77 pregnancies available for first-trimester analysis.

Information on the infants' date of birth, sex, birthweight and presence of one or multiple congenital anomalies was obtained from medical records. The gestational age at birth was calculated from the dating procedure used in the first trimester.

Embryonic growth was studied using CRL measurements performed between 6+0 and 12+6 weeks.

Maternal characteristics were summarised for the total group and stratified by RBC folate quartiles. Distribution across quartiles was tested using Students t-test for normal distributions, Kruskal–Wallis test for non-parametric distributions and chi-squared or exact tests for categorical data depending on the number of cells with an expected value below five. Birthweight was compared between groups by taking into account gestational age at birth using linear regression.

Potential confounders were identified using analysis of variance (ANOVA) with ethnicity and education as explanatory variables, and by calculating Spearman correlation coefficients for the other maternal characteristics listed in Table 1.

Table 1. Ultrasound scans and crown–rump length (CRL) measurements obtained in each gestational week
Gestational weekaNumber of ultrasound scansNumber of CRL measurements
n (2 scans)b % c n (2 scans)b % c % d
  1. a

    Gestational week defined as week+0 to week+6 (week 6 = 6+0 to 6+6 weeks of gestation).

  2. b

    In parentheses, number of pregnancies with two ultrasound scans within the same gestational week.

  3. c

    Percentage of pregnancies with at least one ultrasound scan/CRL measurement (ntotal = 77).

  4. d

    Success percentage of CRL measurements.

654 (1)68.838 (1)48.170.4
768 (1)87.059 (1)75.386.8
875 (1)96.172 (1)92.296.0
974 (2)93.571 (2)89.695.9
1069 (2)87.067 (2)84.497.1
1176 (4)93.572 (4)88.394.7
1268 (0)88.361 (0)79.289.7

To assess the association between maternal RBC folate levels and embryonic growth trajectories, we performed multivariable linear mixed model analyses. By using a mixed model, we take into account that there is a correlation between the observations that belong to the same pregnancy. RBC folate levels were divided into quartiles. Square root transformation of CRL data resulted in linearity with gestational age and a constant variance independent of gestational age, and was therefore used in the analysis. First, we performed a univariate analysis in which we adjusted for gestational age only, and tested for time interaction of RBC folate. In the second or fully adjusted model, we additionally entered fetal sex and all covariates that were significantly correlated with RBC folate levels with and without time interaction into the model. The third and final model was derived from the fully adjusted model after stepwise elimination of all covariates with P values above the 20th percentile.

Analyses were performed in the total group and repeated in a subgroup restricted to pregnancies with the most reliable gestational age, defined as those pregnancies dated on a strictly regular menstrual cycle of 28 ± 3 days and a certain LMP or conception date. In addition, we repeated the analyses in the subgroups of IVF/ICSI pregnancies only, in spontaneous pregnancies only and in spontaneous pregnancies with a reliable gestational age based on a strictly regular menstrual cycle of 28 ± 3 days only.

Linear mixed model analyses were performed using PROC MIXED in SAS software version 9.2 (SAS Institute Inc., Cary, NC, USA). All other analyses were performed using IBM SPSS Statistics Version 20.0 for Windows software (IBM, Armonk, NY, USA).

Results

The median gestational age at the first ultrasound scan was 6+5 weeks [range, 6+0–9+1 weeks; interquartile range (IQR), 6+3–7+0 weeks], and the median number of visits per pregnancy was 6 (range, 4–7; IQR, 6–7). From a total of 484 datasets, 440 (90.9%) were of sufficient quality to perform CRL measurements (Table 1). We performed a median of six (range, 3–7) CRL measurements per pregnancy.

Maternal and pregnancy characteristics are shown in Table 2. Mean maternal age was 32.7 years [standard deviation (SD), 4.5 years] and women predominantly had a high education (60.0%) and were of Dutch descent (78.9%). In 63 (81.8%) pregnancies, gestational age was based on a regular menstrual period of 28 ± 3 days or conception date, including 27 (35.1% of 77 included pregnancies) pregnancies that were conceived after IVF/ICSI treatment. Pregnancy complications occurred in 11 (14.3%) pregnancies. Maternal characteristics were not significantly different across RBC folate quartiles, with the exception of periconception smoking, which was more common in lower quartiles (P = 0.24; Table 2).

Table 2. General characteristics of the study population for the total group and stratified by red blood cell (RBC) folate quartiles
 All (n = 77)MissingRBC folate in quartiles (nmol/l) P
Q1 (814–1223)Q2 (1224–1512)Q3 (1513–1812)Q4 (1813–2696)
  1. BMI, body-mass index; RBC, red blood cell; SGA, small for gestational age; SD, standard deviation.

  2. Data are presented as the median (range) or n (%) unless otherwise specified.

  3. a

    Defined as gestational age based on a menstrual cycle of 28 ± 3 days or conception date.

  4. b

    Defined as weight under the tenth percentile for gestational age, sex and parity according to Dutch reference charts.20

  5. c

    Adjusted for gestational age at delivery.

Maternal (at enrolment)
Maternal age, years (mean ± SD)32.7 ± 4.5331.4 ± 3.632.2 ± 4.433.9 ± 4.933.4 ± 4.10.357
Ethnicity
Dutch60 (78.9%) 13 (68.4)16 (80.0)14 (77.8)17 (89.5)0.331
Other western9 (11.8%)5 (26.3)2 (10.0)1 (5.6)1 (5.3)
Non-western7 (9.2%)1 (5.3)2 (10.0)3 (16.7)1 (5.3)
Education
Low7 (9.3%)22 (10.5)2 (10.5)2 (11.1)1 (5.3)0.951
Middle23 (30.7%)4 (21.1)6 (31.6)6 (33.3)7 (36.8)
High45 (60.0%)13 (68.4)11 (57.9)10 (55.6)11 (57.9)
BMI, kg/m223.3 (18.6–38.3)124.9 (20.7–30.5)24.4 (19.5–31.0)22.6 (18.6–29.7)23.5 (19.4–38.3)0.053
Primiparous49 (63.6%)012 (63.2)13 (65.0)11 (57.9)13 (68.4)0.924
Periconception smoking13 (16.9%)07 (36.8)4 (20.0)1 (5.3)1 (5.3)0.024
Pregnancy and outcome
Conception through IVF/ICSI27 (35.1%)06 (31.6)8 (40.0)6 (31.6)7 (36.8)0.931
Reliable gestational agea63 (81.8%)014 (73.7)17 (85.0)16 (84.2)16 (84.2)0.817
Gestational age at RBC folate determination, week+d7+4 (4+1–11+0)07+2 (4+1–10+3)7+4 (5+2–10+0)7+4 (5+5–11+0)8+2 (6+4–10+6)0.169
Infant sex male36 (46.8%)011 (57.9)10 (50.0)7 (36.8)8 (42.1)0.584
Birth weight, g (mean ± SD)3327 ± 48403157 ± 5593576 ± 3813349 ± 4323214 ± 4720.187c
Gestational age at delivery, week+d39+2 (34+2–41+3)039+4 (35+4–41+2)40+0 (37+6–41+3)39+2 (37+1–41+3)38+6 (34+2–41+1)0.071
Complications 11 (14.3%)03 (15.8)2 (10.0)4 (21.1)2 (10.5)0.772
Maternal 5 (6.5%)0 (0.0)2 (10.0)2 (10.5)1 (5.3%)0.747
Hypertensive complication3 (3.9%)0 (0.0)1 (5.0)1 (5.3)1 (5.3)1.00
Gestational diabetes2 (2.6%)0 (0.0)1 (5.0)1 (5.3)0 (0.0)1.00
Fetal 6 (7.8%)3 (15.8)0 (0.0)2 (10.5)1 (5.3)0.232
Low birthweight (less than 2500 g)4 (5.2%)2 (10.5)0 (0.0)1 (5.3)1 (5.3)0.458
Premature delivery (before 37 weeks)4 (5.2%)2 (10.5)0 (0.0)0 (0.0)2 (10.5)0.250
SGAb4 (5.2%)2 (10.5)0 (0.0)2 (10.5)0 (0.0)0.250

Maternal RBC folate levels were significantly correlated with body mass index (r = −0.26, P = 0.02) and periconception smoking [smokers: mean, 1257 nmol/l (SD, 239 nmol/l); non-smokers: mean, 1627 nmol/l (SD, 475 nmol/l); P < 0.01].

Testing for time interaction showed no significant interaction of RBC folate with gestational age (P = 0.94), and we therefore assumed a linear relation between RBC folate and embryonic size for the remainder of the analyses. The effect estimates from the linear mixed model analyses are displayed in Table 3. Univariate linear mixed model analysis showed that RBC folate in the third quartile (1513–1812 nmol/l) was associated with significantly increased embryonic growth compared with all other quartiles (Poverall = 0.02), including the highest quartile (1813–2969 nmol/l; PQ3–Q4 < 0.01). The estimates for the first two and highest RBC folate quartiles did not differ significantly from each other (P values not shown). Results from the fully adjusted model and the final model derived after backward elimination showed effects of comparable size and significance (Table 3). In Figure 1, regression lines for the two upper RBC folate quartiles derived from the final model are displayed using square root-transformed CRL and after retransformation to the original scale. Compared with the third quartile, RBC folate in the highest quartile was associated with a 1.1-mm (23.5%) and 4.5-mm (7.4%) smaller embryo at 6+0 and 12+0 weeks of gestation, respectively. Estimated differences between RBC folate in the two lowest quartiles and the third quartile were comparable, with embryos that were 1.1 mm (19.4%) and 1.0 mm (17.6%) smaller at 6+0 weeks and 3.6 mm (6.0%) and 3.2 mm (5.4%) smaller at 12+0 weeks of gestation, respectively.

Table 3. Effect estimates of maternal red blood cell (RBC) folate levels in quartiles for crown–rump length (CRL) from the univariate, fully adjusted and final models, using square root-transformed CRL.
Model n subjects n observations Effect estimate (95% CI), √mm P P overall
  1. CI, confidence interval; Poverall, P value for RBC folate in quartiles.

  2. a

    Adjusted for gestational age.

  3. b

    Adjusted for gestational age, periconception smoking, body mass index and fetal sex with and without interaction with gestational age.

  4. c

    Derived from the fully adjusted model after stepwise elimination, adjusted for gestational age and fetal sex.

Univariate a
Q177440−0.26 (−0.45, −0.05)0.010.02
Q2−0.22 (−0.42, −0.03)0.03
Q30 [Reference]
Q4−0.30 (−0.50, −0.10)<0.01
Fully adjusted b
Q176437−0.24 (−0.46, −0.02)0.030.03
Q2−0.24 (−0.45, −0.03)0.02
Q30 [Reference]
Q4−0.30 (−0.50, −0.10)<0.01
Final c
Q177440−0.24 (−0.44, −0.04)0.020.02
Q2−0.21 (−0.41, −0.02)0.03
Q30 [Reference]
Q4−0.29 (−0.49, −0.09)<0.01
Figure 1.

Regression lines for crown–rump length (CRL) growth conditional on red blood cell (RBC) folate quartiles derived from the final model adjusted for fetal sex, displayed using square root CRL data (A) and after retransformation to the original CRL scale (B). Regression lines are shown for the upper two quartiles only, as the regression lines for both lower quartiles are very close to those of the upper quartile.

We repeated the analysis restricted to specific subgroups of pregnancies (Table 4). In the subgroups of pregnancies with a reliable gestational age (i.e. IVF/ICSI pregnancies or spontaneous pregnancies with a regular menstrual cycle of 28 ± 3 days), spontaneous pregnancies and spontaneous pregnancies with a regular menstrual cycle of 28 ± 3 days only, effect estimates and significance from the univariate model were similar to those observed in the total group, with the exception that, in the latter group, the overall P value and the P value of the first quartile were no longer significant (Poverall = 0.07, PQ1 = 0.11). In the subgroup of IVF/ICSI pregnancies, effect estimates pointed in the same direction, but were smaller and did not reach significance, except for the first quartile, presumably because of the small numbers (Poverall = 0.21, PQ1/2/4-Q3 = 0.04/0.37/0.14).

Table 4. Effect estimates of maternal red blood cell (RBC) folate levels in quartiles for crown–rump length from the univariate models for different subgroups of pregnancies, using square root-transformed CRL and adjusted for gestational age
  n subjects n observations Effect estimate (95% CI), √mm P P overall
  1. CI, confidence interval; Poverall, P value for RBC folate in quartiles.

  2. a

    Strictly regular menstrual cycle of 28 ± 3 days.

IVF/ICSI or spontaneous regular a
Q163367−0.23 (−0.43, −0.03)0.030.03
Q2−0.22 (−0.41, −0.03)0.03
Q30 [Reference]
Q4−0.28 (−0.47, −0.08)<0.01
IVF/ICSI
Q127163−0.18 (−0.35, −0.01)0.040.21
Q2  −0.07 (−0.23, 0.09)0.02
Q3  0 [Reference]
Q4  −0.13 (−0.29, 0.04)0.14
Spontaneous      
Q150277−0.29 (−0.57, −0.01)0.040.04
Q2−0.32 (−0.60, −0.03)0.03
Q30 [Reference]
Q4−0.39 (−0.68, −0.10)<0.01 
Spontaneous regular a
Q135197−0.27 (−0.59, −0.05)0.110.07
Q2−0.35 (−0.66, −0.03)0.03
Q30 [Reference]
Q4−0.39 (−0.71, −0.06)0.02

Discussion

Main findings

Maternal first-trimester RBC folate appears to follow an optimum curve in which both lower (<50th percentile, 814–1513 nmol/l) and very high levels (>75th percentile, 1813–2936 nmol/l) are associated with reduced embryonic size.

Strengths and weaknesses

In this first study on maternal RBC folate and embryonic growth trajectories, CRL measurements were of excellent quality as a result of the use of 3D ultrasound scans with a virtual reality environment.[18] High precision was achieved using the means of measurements performed in triplicate. Furthermore, the assessment of long-term RBC folate reflects the periconception maternal folate status.

We excluded pregnancies with major congenital malformations and pregnancies resulting in fetal or neonatal demise. Compared with the general population, our study population was well educated, more often conceived using IVF/ICSI, more often used FA supplements and was likely to be at a higher risk for pregnancy complications. These factors may explain the overall high RBC folate levels and absence of folate deficiencies. Future research must elucidate whether the observed association also applies to the general population and whether it is also associated with pregnancy outcome. We are aware that the inclusion of both spontaneously conceived pregnancies and IVF/ICSI pregnancies may have decreased the precision of the determination of gestational age. We therefore excluded pregnancies with a difference of more than 6 days between pregnancy dating using the LMP compared with CRL. Although we cannot exclude the possibility that folate status influences menstrual cycle regularity and endometrial receptivity important for implantation, the direction of misdating is likely to be random and randomly distributed across RBC folate quartiles. This is supported by the results from the subgroup analyses restricted to pregnancies with the most reliable gestational age. Our numbers were too small to observe significant results in IVF/ICSI pregnancies only, although effect estimates point in the same direction.

Interpretation

Our results suggest a difference in embryonic growth; however, we were unable to demonstrate a significant interaction between RBC folate and gestational age on the square root scale. Results are therefore discussed as a difference in embryonic size rather than embryonic growth.

Our data support the results from animal studies in which high folate levels have been investigated. In rodents, a 20-fold enriched folate diet was associated with a smaller embryo and decreased newborn length and weight, but also with an increase in embryonic development at 10.5 days post coitum.[21, 22] However, in these studies, synthetic FA and plasma folate were studied compared with long-term RBC folate in our study.

In previous human studies, associations between first-trimester RBC folate and prenatal growth have been assessed at birth only. Results show positive associations with birthweight and head circumference.[9-12] This substantiates our positive association between embryonic size and RBC folate up to 1812 nmol/l, as the previously reported levels of first-trimester RBC folate were below this level. Only Takimoto et al.[23] have reported a high first-trimester mean RBC folate of 1317 nmol/l (SD, 824 nmol/l), but they observed no associations with birthweight or head circumference, which may be because of the small sample size (n = 51) and the performance of continuous rather than stratified analysis for different cut-offs for RBC folate.

Mechanisms by which very high maternal periconception RBC folate could affect embryonic size remain to be elucidated. The periconception period is highly important with regard to cell multiplication, differentiation and epigenetic programming of the embryo and placenta by DNA methylation of genes implicated in growth and development.[24] One-carbon metabolism provides one-carbon groups for these processes, of which folate is an important substrate. High folate levels also require high levels of cofactors, such as other B vitamins, a shortage of which can also derange these biological processes implicated in embryonic growth. We have shown previously that maternal periconception FA supplementation is associated with increased DNA methylation of the maternally imprinted embryonic growth gene insulin-like growth factor-2 (IGF2) in the very young child.[25] In addition, increased methylation of IGF2 was associated with a lower birthweight. This epigenetic effect of FA supplementation is in line with the results of another study, showing that periconception exposure to famine is associated with decreased IGF2 methylation.[26] These findings are further supported by a study in mice demonstrating widespread hypomethylation and adverse effects on health in later life after preconception exposure to a folate-deficient diet.[27] Thus, increased periconception RBC folate may lead to increased DNA methylation and, consequently, silencing of imprinted genes implicated in early prenatal growth with long-term health consequences.

Another potential mechanism stems from the observation that high folate levels can inhibit folate-dependent enzymes.[28, 29] In human intestinal and renal epithelial cells, long-term excessive FA supplementation leads to a specific and significant down-regulation of folate uptake.[30] FA supplementation has been associated with elevated folate levels in amniotic fluid.[31] In late first trimester, the maternal-to-fetal exchange of nutrients begins and high folate levels reach the fetus and placenta. Umbilical folate concentrations exceed maternal concentrations[32] and, although investigated at birth only, embryonic concentrations may well begin to rise towards the end of the first trimester, thus potentially leading to the down-regulation of folate-dependent enzymes and uptake, resulting in the inhibition of DNA synthesis and growth.

Finally, the high RBC folate levels in our study population are probably caused by long-term FA supplement intake. FA in dosages above 200 μg cannot completely be reduced and bound to proteins in the plasma and thus enters the blood plasma in the unmetabolised form.[33] Although unmetabolised FA does not appear to accumulate in the fetus,[34] the effects of repeated exposure over a substantial period of time remain unclear.

The mean absolute embryonic sizes in our study population were comparable with data from Pexsters et al.[35] and Robinson and Fleming,[19] except for size at 6 weeks, where our embryos appeared to be larger than the data reported by Pexsters et al.[35] [mean, 4.6 mm (95% reference interval, 2.2, 7.9) versus mean, 1.9 mm (95% reference interval, 0.4, 4.5); Supporting information Table S1). This difference may be explained by differences in precision of pregnancy dating and measurements using 3D ultrasound and virtual reality.

Although we are often inclined to view embryonic size as the larger the better, the consequences of increased embryonic size have not yet been clarified. Embryonic growth has been positively associated with birthweight[36-39] and, as birthweight is an important determinant for health in later life, this indeed suggests a positive quality. However, long-term consequences of potential epigenetic modifications and the determination of the optimum of RBC folate with regard to embryonic growth are issues still to be unravelled. Given the vast amount of influences on prenatal growth in the second and third trimesters, our study population is too small to assess the implications of the association between RBC folate and embryonic size for subsequent fetal growth and pregnancy outcome. Finally, because the implications of a smaller embryo are unclear and high-dose FA supplementation is unequivocally effective in the prevention of NTD recurrences,[40-42] our results must be interpreted with caution.

Conclusions

Periconceptional maternal RBC folate levels above 906 nmol/l prevent NTDs in the offspring.[43] In our study, 95% of all women demonstrated RBC folate levels exceeding this threshold. Beneficial and harmful effects of the long-term use of high doses of FA supplements resulting in very high periconception RBC folate levels have not yet been clarified,[44] whereas, with the increasing use of FA, multivitamins and fortified foods, its safety is becoming increasingly important. Therefore, more research is needed on the optimal folate level with regard to embryonic size and growth, including effects on subsequent fetal growth and pregnancy outcome, with potential implications for current folate recommendations.

Disclosure of interests

The authors declare no conflicts of interest.

Contribution to authorship

EMvU analysed the data and wrote the first draft of the manuscript. SvG contributed to data acquisition and logistics. PHCE and SPW assisted in data analysis and the interpretation of the results. AHJK and NE supervised the CRL measurements from the ultrasound scans. JL, head of the laboratory, supervised the RBC folate measurements. JSEL and EAPS were responsible for the included patients and the infrastructure of the study. RPMS-T initiated and designed the study, supervised all aspects of the study and contributed to all versions of the manuscript. All authors were involved in the interpretation of the results and revision of the manuscript, and approved the final version.

Details of ethics approval

This study protocol has been ethically approved by the Medical Ethical Committee of the Erasmus University Medical Centre.

Funding

This research was funded by the Department of Obstetrics and Gynaecology, Erasmus University Medical Centre, Rotterdam, the Netherlands.

Acknowledgements

We are grateful to all participating women in this study and to the Rotterdam Predict study team for recruitment, data acquisition and performance of weekly ultrasounds and I-Space measurements. We thank Mark Wildhagen, PhD, for his contribution to data management.

Ancillary