Maternal lifestyle risk factors and human embryonic morphologic development using 3D ultrasound: a population-based prospective cohort study

Abstract

STUDY QUESTION

Are periconception maternal modifiable lifestyle factors (i.e. smoking, alcohol consumption, caffeine intake, folic acid supplement use, and physical activity) associated with embryonic morphologic development and growth in a large population-based cohort study?

SUMMARY ANSWER

Inadequate folic acid supplement use and insufficient serum folate levels were associated with delayed embryonic morphologic development and reduced growth, while no associations were observed for smoking, alcohol, caffeine intake, or physical activity.

WHAT IS KNOWN ALREADY

Modifiable maternal lifestyle factors during the periconception period may influence embryonic growth and development. Prior studies suggest associations between smoking, folic acid use, and first trimester growth, but evidence on morphologic development remains limited.

STUDY DESIGN, SIZE, DURATION

This study was embedded in the Generation R Next Study, a population-based periconception cohort in Rotterdam, the Netherlands, conducted between 2017 and 2021. Of the 2173 included spontaneous singleton pregnancies, 993 had reliable pregnancy dating, 3D ultrasound data up to 10 + 2 weeks of gestation, and available information on periconception lifestyle risk factors.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Embryonic development and growth were assessed in vivo at 7 and 9 weeks of gestation using 3D transvaginal ultrasound and virtual reality techniques. Morphologic development was determined using the Carnegie staging system based on limb position, embryonic curvature, and brain morphology; crown-rump length (CRL) was used to assess growth. Periconception lifestyle factors, including smoking, alcohol and caffeine intake, physical activity, and folic acid supplement use, were assessed using early pregnancy questionnaires. Linear mixed models were used to examine the association between lifestyle factors and embryonic Carnegie stages and CRL. Analyses were adjusted for gestational age at ultrasound, maternal age, ethnicity, education level, parity, fetal sex, and the other lifestyle factors.

MAIN RESULTS AND THE ROLE OF CHANCE

Inadequate folic acid supplement use was associated with a 0.51-day delay in embryonic development and a 0.72-mm reduction in CRL (−1.6%) at 10 + 2 weeks of gestation. Insufficient serum folate levels (<25.5 nmol/l) were associated with a 1.01-day developmental delay and a 1.24-mm reduction in CRL (−3.7%). No significant associations were observed for smoking, alcohol or caffeine intake, or physical activity. Results remained robust after adjustment for confounders.

LIMITATIONS, REASONS FOR CAUTION

Lifestyle exposures were self-reported, which may have introduced recall or social desirability bias. Gestational age determination was based on the last menstrual period, which can result in dating discrepancies due to variability in cycle length or inaccurate recall, but strict inclusion criteria were applied to minimize misclassification. The relatively healthy study population could limit the generalizability of our results to higher-risk groups.

WIDER IMPLICATIONS OF THE FINDINGS

Our findings emphasize the importance of initiating folic acid supplementation in the preconception period to ensure sufficient maternal folate levels, which may support optimal embryonic morphologic development and growth during the first trimester. These results reinforce public health recommendations advocating timely folic acid supplementation to promote healthy early embryonic development.

STUDY FUNDING/COMPETING INTEREST(S)

The Generation R Next Study is financially supported by the Erasmus University Medical Centre Rotterdam, the Erasmus University Rotterdam, and the Netherlands Organization for Health Research and Development. R.G. received funding from the Netherlands Organization for Health Research and Development (NWO, ZonMw VIDI 09150172110034, and NWO, ZonMW, grant number 05430052110007) and an European Research Council Starting Grant (ERC-2024-STG 101161004). V.W.V.J. received a grant from the Netherlands Organization for Health Research and Development (NWO, ZonMw 05430052110007) and a European Research Council Consolidator Grant (ERC-2014-CoG-648916). The authors declare no conflicts of interest.

TRIAL REGISTRATION NUMBER

N/A.

Introduction

The increasing prevalence of poor nutrition and lifestyle choices poses a significant public health concern. Modifiable lifestyle factors including smoking, alcohol consumption, and lack of physical activity are major contributors to chronic diseases like obesity and type 2 diabetes, cardiovascular conditions, and certain cancers (Chiuve et al., 2008; Loef and Walach, 2012; Altobelli et al., 2020; Li et al., 2020; Oster and Chaves, 2023). This issue is particularly concerning for women of reproductive age, because poor lifestyle choices pose additional risks specifically in pregnancy, impacting fetal growth, development, and the long-term health of the child (Barker, 1999, 2007; Fleming et al., 2018).

Emerging evidence suggests that lifestyle factors in the periconception period, from 14 weeks before conception up to and including the first 10 weeks of gestation, can influence embryonic size and growth, measured in crown-rump length (CRL), and subsequent birth outcomes and childhood health (Mook-Kanamori et al., 2010; van Uitert et al., 2013; Jaddoe et al., 2014; Van Dijk et al., 2018; Oostingh et al., 2019). A growing body of evidence suggests that smoking is associated with a smaller CRL compared with non-smoking women (Mook-Kanamori et al., 2010; Prabhu et al., 2010; van Uitert et al., 2013; Van Dijk et al., 2018; Stamatopoulos et al., 2022), whereas preconceptional initiation of folic acid supplement use has been associated with increased CRL compared with postconceptional initiation or no supplement use (Mook-Kanamori et al., 2010; van Uitert et al., 2013; Van Dijk et al., 2018). Evidence regarding the association between alcohol consumption, caffeine consumption, and physical activity and embryonic growth remains unclear (Mills et al., 1993; Bakker et al., 2010; van Uitert et al., 2013; Van Dijk et al., 2018; van der Windt et al., 2024).

Recent advancements in high-resolution 3-dimensional (3D) ultrasound combined with innovative virtual reality (VR) techniques, now offer real depth perception and a more sensitive assessment of embryonic size (including CRL) and morphologic development. The century-old Carnegie staging system, a measure of human morphologic development, divides the embryonic period into 23 stages based on both internal and external characteristics. Originally developed based on ex vivo human embryos, 3D ultrasound in combination with VR now also allows in vivo assessment (Rousian et al., 2018). This comprehensive approach considers the entire embryo, rather than just size alone, and therefore potentially provides a more accurate proxy for embryonic health. Only within two previously published studies, embryonic morphology was assessed in relation to periconceptional smoking and levels of red blood cell (RBC) folate, an indicator of folic acid supplement use and dietary intake. The studies were conducted within the Rotterdam Periconception Cohort, a tertiary hospital-based cohort study. These studies concluded that both periconceptional smoking and low RBC folate levels were associated with delayed embryonic development (Parisi et al., 2017; Pietersma et al., 2022).

Against this background, we aim to investigate associations between periconception maternal modifiable lifestyle factors (i.e. smoking, alcohol consumption, caffeine intake, folic acid supplement use, and physical activity) and embryonic morphologic development and growth, in a large population-based cohort study.

Materials and methods

Study design

This study was embedded in a subgroup of the Generation R Next Study, a population-based prospective cohort study from the preconception period onward in Rotterdam, the Netherlands, and is part of the Generation R Study Programme (Kooijman et al., 2016; Boxem et al., 2024). Its primary aim is to identify preconception and early-pregnancy determinants of fertility, embryonic development, and childhood outcomes. Eligible participants included women and their partners aged ≥18 years, residing in Rotterdam, and either actively trying to conceive or already pregnant, between 2017 and 2021. While enrollment aimed for inclusion during preconception or early pregnancy, enrollment was allowed until delivery. In total, 33.2% and 52.8% of all inclusions were during the preconception period or in the first trimester of pregnancy, respectively. By the end of the study period, 4033 participant episodes from 3602 unique women were included, leading to 3575 pregnancies of 3197 unique women. Study approval was obtained by the Medical Ethical Committee of the Erasmus Medical Center, University Medical Center, Rotterdam (MEC 2016-589, NL57828.078.16). Written informed consent was obtained from all participating women. Findings were reported following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline (von Elm et al., 2007).

Study population

Although enrollment into the overall Generation R Next cohort was possible until delivery, the current analyses focused exclusively on early pregnancy participation. Therefore, only participants with spontaneous, singleton pregnancies who were recruited preconceptionally or during the first trimester, up to 10 + 2 weeks of gestation, which marks the end of the embryonic period (O’Rahilly and Müller, 1987, 2010), were eligible for inclusion. This resulted in a total of 2173 pregnancies. Miscarriages, terminated pregnancies, and pregnancies resulting in perinatal death were excluded (n = 368). Gestational age was calculated from the first day of the last menstrual period (LMP), as CRL, which is used to estimate gestational age in clinical practice, denominated a primary outcome of this study, and doesn’t reflect embryonic growth variation. Pregnancies were excluded if they had an unknown LMP, an irregular cycle (<21 and >35 days), or if there was a difference of more than 6 days between the gestational age based on LMP and the gestational age based on CRL (n = 612). Pregnancies were excluded if participants moved out of the study area, withdrew consent or were lost to follow-up (n = 78). Pregnancies were also excluded if both 3D Carnegie stage and 3D CRL measurement were missing between 6 + 0 and 10 + 2 weeks of gestation (n = 121). This resulted in a total study population of 993 pregnancies, recruited either preconceptionally (29.6%) or in the first trimester of pregnancy (70.4%) (Fig. 1).

Figure 1.

Flowchart of participants included in the study. LMP, last menstrual period; CRL, crown-rump length.

Lifestyle risk factors

The included maternal modifiable lifestyle risk factors (smoking, alcohol consumption, caffeine intake, folic acid supplement use, and physical activity) were selected based on previous research (Graafland et al., 2025) and were assessed using questionnaires administered in early pregnancy.

Periconceptional smoking

Women reported on smoking during preconception (3 months before pregnancy) and during pregnancy in the first trimester. Additional information was collected on whether smoking was discontinued and, if so, when during the periconceptional period. For analysis, periconceptional smoking was categorized as no (no smoking or smoking discontinued within 3 months before conception) or yes (smoking during pregnancy). Women who discontinued smoking after conception but within the first trimester were classified as having smoked during pregnancy, reflecting smoking exposure at or around the time of conception.

Periconceptional alcohol consumption

Women reported alcohol use during preconception and during pregnancy in the first trimester. Response options included no alcohol use, alcohol use that was discontinued within the 3 months before conception, or alcohol use during pregnancy. For analysis, alcohol consumption was categorized as no (no alcohol use or alcohol use only preconceptionally) or yes (alcohol use during pregnancy), reflecting alcohol exposure at or around the time of conception.

Caffeine consumption

Women who reported caffeine intake in pregnancy were asked to specify the type of caffeinated beverages consumed (coffee, tea, cola, or energy drinks) and the daily amount. Total intake was calculated using standard reference values for caffeine content in the Netherlands (Voedingscentrum, 2022). One unit was defined as ∼60 mg of caffeine, corresponding to one standard cup of coffee. Caffeine consumption was subsequently categorized as none, <2 units/day, 2–3.9 units/day, or ≥4 units/day.

Physical activity

Participants were asked to report the amount of time spent per week in moderate and strenuous physical activity. Moderate physical activity was defined in the questionnaire as activities during which participants were generally still able to talk, but with increased heart rate and breathing (e.g. brisk walking, cycling at <20 km/h, or leisurely swimming). Strenuous physical activity was defined as activities during which participants were unable to speak more than a few words consecutively due to markedly increased breathing and heart rate (e.g. running, cycling at >20 km/h, fast swimming, or intensive gardening). Physical activity was then categorized as not active, moderately active (<150 min of moderate activity per week or <75 min of strenuous activity per week), or very active (>150 min of moderate activity per week or >75 min of strenuous activity per week). A very active lifestyle was used as the reference category, as maintaining an active lifestyle is recommended during the periconception period (ACOG, 2020).

Folic acid supplement use

In accordance with the recommendation to initiate folic acid supplement use preconceptionally, we categorized folic acid supplement use in two groups: adequate use (preconceptional initiation of supplement use) and inadequate use (postconceptional initiation of supplement use or no use at all) (Centers for Disease Control, 1993; World Health Organization, 2007).

Maternal serum folate concentrations

As a key biomarker complementing folic acid supplement use, serum folate concentrations were also assessed. Maternal serum samples were collected during the first prenatal visit in early pregnancy (median 7 + 5 weeks; 90% range 6 + 5 to 9 + 2 weeks). Folate concentrations in these samples were determined using an immunoelectrochemoluminescence assay. The upper limit of detection (LOD) of folate was 45.4 nmol/l and any concentration above this limit was imputed as 45.4 nmol/l. Serum folate concentrations were categorized as sufficient (≥25.5 nmol/l) or insufficient (<25.5 nmol/l), based on recommendations for women of reproductive age (Cordero et al., 2015; Chen et al., 2019).

Outcomes

During the first trimester, transvaginal 3D ultrasound volume sweeps of the embryo were performed by trained sonographers following cohort-specific guidelines (Campbell and Wilkin, 1975; Verburg et al., 2008; Ioannou et al., 2013; Bilardo et al., 2023). Sonographers were trained following national standards set by the Dutch Ministry of Public Health, Welfare, and Sports; thus, measurement variability was assumed to be representative of clinical practice in the Netherlands. Ultrasound examinations were carried out using the Voluson© E10 (GE Healthcare), Philips© Affiniti 70, or Philips© Epiq 7 systems. Transvaginal transducers with varying frequency ranges (4.0–9.0 MHz or 4.5–11.9 MHz) were selected according to the study protocol, primarily based on gestational age and optimal visualization of embryonic structures, with higher-frequency probes more commonly used in earlier gestations. If initial image quality was insufficient, the 3D volume sweep was repeated using an alternative transducer. Routine quality control procedures were performed to monitor image quality throughout the study period. At each ultrasound examination, two replicate 3D embryonic volume sweeps were acquired to ensure adequate image quality for subsequent analysis.

To enable offline analysis, all 3D ultrasound data were converted into Cartesian (rectangular) volumes. In-house developed V-Scope software (Erasmus MC, Rotterdam, the Netherlands) (Koning et al., 2009) was used to generate embryonic holograms on a VR system named the I-Wall (Rousian et al., 2018). This is a flat 3D rear-projection-based semi-immersive VR system, offering real depth perception using stereoscopic glasses. For this study, embryonic morphological development and embryonic growth were assessed from the offline 3D ultrasound data acquired during the embryonic period, defined as gestational ages ≤10 + 2 weeks. Therefore, although ultrasound examinations were scheduled at ∼7, 9, and 11 weeks of gestation, only scans within this gestational age window were included in the analyses. Carnegie stages were determined in 967 pregnancies, with 50.4% (488 pregnancies) assessed at two timepoints. Similarly, 3D CRL measurements were performed in 991 pregnancies, with 60.9% (604 pregnancies) evaluated at two timepoints. The availability of one or two scans during the embryonic period varied due to later inclusion in the study, missed appointments, gestational age at the time of the ultrasound exceeding the predefined study period (≥10 + 3 weeks), participant preference not to undergo transvaginal 3D ultrasound and insufficient image quality to allow reliable offline measurements.

Embryonic morphologic development was assessed using the Carnegie staging system. This system was originally designed based on ex vivo human embryos, using internal, external, and histologic criteria. VR enables the in vivo determination of these stages, based on the following criteria: position of upper and lower limbs, curvature of the embryo, and morphological features of the pros-, mes-, and rhombencephalon (O’Rahilly and Müller, 2010). Morphologic development was visually compared to the Carnegie stages 13–23, as refined by O’Rahilly and Müller in 1987 (O’Rahilly and Müller, 1987, 2010; Verwoerd-Dikkeboom et al., 2008a; Rousian et al., 2018; Flierman et al., 2023; Pietersma et al., 2023). Figure 2 shows representative examples of embryos at Carnegie stages 15, 17, 20, and 23. Before initiating measurements, assessors completed a structured training and learning period. Measurements were only performed after excellent reproducibility was achieved (ICC > 0.90). Embryonic growth was assessed using CRL measurements, the length from the cranium to the caudal rump in a straight line within the mid-sagittal plane (Robinson, 1973; Robinson and Fleming, 1975). Using VR techniques, very precise caliper positioning is possible (Verwoerd-Dikkeboom et al., 2008b).

Figure 2.

Comparison of embryos at four different Carnegie stages from the original Carnegie collection and in vivo embryos assessed using 3D ultrasound and virtual reality. Images are shown in the dorsal, sagittal, and midsagittal planes. The numbers in the left-hand panels indicate the Carnegie stages, assigned based on the assessment of internal and external morphological features of the embryo, including the development and position of the limbs, curvature of the embryo, and morphological features of the prosencephalon, mesencephalon, and rhombencephalon.

Covariates

Information on maternal age at conception (years), ethnicity (Dutch; non-Dutch, European; non-Dutch, non-European), educational level (low: no or primary education; medium: secondary education; high: university degree), and parity (nulliparous; parous) was obtained through questionnaires at enrollment. Fetal sex and birth weight were obtained from midwifery registries and obstetric medical records.

Statistical analyses

First, we conducted a nonresponse analysis, comparing participants that were included in our study to those that were excluded due to an unknown or an unreliable LMP, or an irregular menstrual cycle by using the t-test, the Mann–Whitney U-test, or the χ² test. To account for selection bias by irregular or unknown cycle, inverse probability weighting was performed, by using information available from both groups (i.e. irregular or unknown cycle and known cycle) to predict the probability of participation in the study. The inverse of those probabilities was used as weights in the analyses to ensure representative results for the initial cohort population (Weuve et al., 2012; Weisskopf et al., 2015). Second, linear mixed models (LMM) were used to examine the associations between each modifiable lifestyle risk factor (i.e. periconception smoking, alcohol consumption, caffeine consumption, physical activity, and folic acid supplement use) and longitudinally measured 3D Carnegie stages and CRL. To improve interpretability, additional analyses were performed for lifestyle factors that showed an association with Carnegie stage or CRL. Linear mixed models including interaction terms between the lifestyle factor and gestational age were fitted. For Carnegie stage analyses, the interaction model was used to estimate the gestational age at which Carnegie stage 23 (representing the end of embryonic development) was reached for exposed and unexposed pregnancies; the difference between these estimates was expressed as a delay in days. For CRL analyses, estimated CRL values for exposed and unexposed pregnancies were derived at a fixed gestational age of 10 + 2 weeks, and differences were expressed as percentage change. In the crude model (model 1), the association between lifestyle risk factors and Carnegie stage and CRL was adjusted only for gestational age. Model 2 included additional adjustments for potential confounders including maternal age at conception, ethnicity, educational level, parity, and fetal sex. To examine the independent associations of each lifestyle factor with the outcomes, model 3 was further adjusted for the other studied lifestyle risk factors. Third, we conducted a separate analysis to assess the association between serum folate levels and trajectories of Carnegie Stages and CRL. Folate levels were evaluated as both a continuous variable and a categorical variable (sufficient (≥25.5 nmol/l) vs insufficient (<25.5 nmol/l)). The same adjustments were applied across for model 1. Model 2 was additionally adjusted for potential confounders, including maternal age at conception, body mass index (BMI), ethnicity, educational level, parity, fetal sex, and smoking. As a sensitivity analysis, we analyzed a subgroup of the population, excluding the participants with a serum folate above the upper LOD (>45.3 nmol/l), to reduce the influence of extreme outliers. Potential confounders for all analyses were selected a priori based on a directed acyclic graph (DAG) and associations observed in existing literature (Supplementary Fig. S1). Missing data for potential confounders were lower than 10%. To limit selection bias and loss of statistical power, multiple imputation was applied. A total of 25 complete datasets were obtained and analyzed using standard procedures for multiple imputation (Graham et al., 2007; Sterne et al., 2009). All statistical analyses were performed using R Statistical Software (version 4.3.0; R Foundation for Statistical Computing, Vienna, Austria).

Results

Characteristics of the study population

A total of 993 women were included in the study, with 294 recruited preconceptionally and 699 during early pregnancy (up to 10 + 2 weeks of gestation). A total of 1597 3D datasets were collected, enabling 1455 measures of Carnegie stage and 1595 measures of CRL. Baseline characteristics of the study population are presented in Table 1. Women had a mean age at conception of 31.4 years (SD: 3.9) and a median pre-pregnancy BMI of 23.2 kg/m² (IQR: 21.2–26.0). Most women were of Dutch nationality, highly educated, and nulliparous. During the periconception period, 69.5% of women used folic acid supplements adequately, and 85.3% had sufficient serum folate levels. Serum folate concentrations exceeded the assay’s upper LOD in 19.5% of women. Overall 22.2% of women reported smoking in pregnancy; of these, 41.5% discontinued smoking during the first trimester. Additionally, 18.3% pregnant women consumed alcohol, and 8.7% consumed more than 4 units of caffeine daily. Furthermore, 75.2% of women were very physically active during the first trimester. Nonresponse analysis revealed that women with a known LMP and a regular cycle were more likely to be of Dutch nationality, highly educated, have higher monthly household incomes, be smokers, consume alcohol, and engage in more physical activity compared to those with an unknown LMP or an irregular cycle (Supplementary Table S1).

Table 1.

Baseline characteristics of the total population.

Maternal characteristics	Distribution	Missing
	N = 993
Maternal age at conception (years)	31.4 (3.9)	0
Pre-pregnancy BMI (kg/m2)	23.2 (21.2–26.0)	64 (6.5)
Ethnicity (%)		25 (2.5)
Dutch	665 (67.7)
Non-Dutch, European	111 (11.5)
Non-Dutch, non-European	202 (20.9)
Maternal educational level (%)		32 (3.2)
Low	34 (3.5)
Medium	201 (20.9)
High	726 (75.6)
Monthly household income (%)		108 (10.9)
<2000 euro	68 (7.7)
2000–4000 euro	285 (32.2)
>4000 euro	532 (60.1)
Parity (%)		45 (4.5)
Nulliparous	627 (66.1)
Parous	321 (33.9)
First trimester smoking (yes, %)	200 (22.2)	90 (9.1)
First trimester alcohol consumption (yes, %)	175 (18.3)	38 (3.8)
Daily caffeine consumption, unitsa (%)	111 (55–167)	71 (7.2)
None	50 (5.4)
<2 units	444 (48.2)
2–3.9 units	348 (37.7)
> 4 units	80 (8.7)
Physical activity (%)		72 (7.3)
Not active	17 (1.9)
Moderately active (<150 min moderate or <75 min strenuous intensity)	211 (22.9)
Very active (>150 min moderate or >75 min strenuous intensity)	693 (75.2)
Folic acid supplement use (inadequate, %)b	287 (30.5)	52 (5.2)
Serum folate (nmol/l)	37.7 (30.1–44.7)	112 (11.3)
Sufficientc (≥25.5 nmol/l)	752 (85.3)
Insufficient (<25.5 nmol/l)	129 (14.7)
Birth outcomes
Fetal sex (male, %)	522 (55.6)	0
Gestational age at birth, weeks (IQR)	39.7 (38.7–40.6)	0
Preterm birth (%)	54 (5.4)	0
Birth weight, Hoftiezer percentile (IQR)	0.42 (0.18–0.67)	22 (2.2)

Values are presented as mean (SD), median (IQR), or number of participants (valid %).

^aOne unit reflects 90 mg caffeine.

^bFolic acid supplement use; adequate: preconceptional initiation of supplement use, inadequate: postconceptional initiation of supplement use or no use at all.

^cSufficient levels ≥25.5 nmol/l serum folate, based on recommendations for women of reproductive age.

Lifestyle risk factors and embryonic development

Table 2 shows the associations between maternal periconception modifiable lifestyle risk factors and morphologic development according to the Carnegie stages. Linear mixed models showed that inadequate folic acid supplement use was associated with lower than expected values of the Carnegie stages, compared with adequate preconceptional supplementation (β −0.24 [95% CI: −0.39 to −0.10]). After adjustment for covariates and other lifestyle risk factors, the associations persisted with comparable effect estimates. Essentially, this indicates that embryo’s from mothers with inadequate folic acid supplementation reach the final Carnegie stage with a delay of 0.51 days compared to those with adequate supplementation. No associations were observed between periconception smoking, alcohol consumption, caffeine consumption, and physical activity and embryonic morphologic development according to the Carnegie stages.

Table 2.

Association between periconceptional lifestyle risk factors and embryonic development (Carnegie stages) and growth (crown-rump length).

	Carnegie stage (beta (95% CI))			Crown-rump length (beta (95% CI))
	Model 1a	Model 2b	Model 3c	Model 1a	Model 2b	Model 3c
Periconception smoking		N = 879			N = 901
No	Reference	Reference	Reference	Reference	Reference	Reference
Yes	−0.06 (−0.28 to 0.14)	−0.06 (−0.27 to 0.15)	−0.03 (−0.24 to 0.19)	−0.19 (−0.75 to 0.36)	−0.19 (−0.48 to 0.10)	−0.16 (−0.46 to 0.14)
Periconception alcohol consumption		N = 929			N = 953
No	Reference	Reference	Reference	Reference	Reference	Reference
Yes	0.06 (−0.11 to 0.23)	0.02 (−0.15 to 0.20)	0.04 (−0.13 to 0.22)	0.20 (−0.29 to 0.68)	−0.05 (−0.29 to 0.18)	−0.01 (−0.25 to 0.22)
Daily caffeine consumption		N = 897			N = 920
None	Reference	Reference	Reference	Reference	Reference	Reference
<2 units	−0.13 (−0.44 to 0.18)	−0.20 (−0.51 to 0.11)	−0.25(−0.56 to 0.07)	−0.25 (−1.12 to 0.62)	−0.13 (−0.56 to 0.30)	−0.17 (−0.60 to 0.26)
2–3.9 units	−0.14 (−0.45 to 0.18)	−0.21 (−0.52 to 0.11)	−0.26 (−0.58 to 0.06)	−0.27 (−1.15 to 0.61)	−0.19 (−0.63 to 0.25)	−0.23 (−0.67 to 0.21)
≥4 units	−0.15 (−0.53 to 0.22)	−0.21 (−0.59 to 0.17)	−0.28 (−0.67 to 0.10)	−0.42 (−1.47 to 0.63)	−0.20 (−0.72 to 0.33)	−0.26 (−0.78 to 0.27)
Phsyical activity		N = 896			N = 919
Not active	0.35 (−0.18 to 0.87)	0.41 (−0.13 to 0.95)	0.38 (−0.16 to 0.93)	1.53 (0.11–2.95)*	0.72 (−0.02 to 1.47)	0.72 (−0.02 to 1.47)
Moderately active	−0.07 (−0.23 to 0.09)	−0.06 (−0.22 to 0.11)	−0.05 (−0.21 to 0.12)	−0.12 (−0.56 to 0.42)	0.14 (−0.08 to 0.37)	0.15 (−0.08 to 0.37)
Very active	Reference	Reference	Reference	Reference	Reference	Reference
Folic acid supplement used		N = 917			N = 939
Adequate	Reference	Reference	Reference	Reference	Reference	Reference
No or inadequate	−0.24 (−0.39 to −0.10)*	−0.24 (−0.39 to −0.08)*	−0.25 (−0.41 to −0.09)*	−0.46 (−0.87 to −0.05)*	−0.23 (−0.44 to −0.02)*	−0.23 (−0.44 to −0.02)*

^aModel 1: GA.

^bModel 2: GA, maternal age at conception, ethnicity, educational level, parity, and fetal sex.

^cModel 3: GA, maternal age at conception, ethnicity, educational level, parity, fetal sex, periconception smoking, alcohol consumption in first trimester, caffeine consumption, physical activity, and folic acid supplement use (unless exposure of interest).

^dFolic acid supplement use; adequate: preconceptional initiation of supplement use, inadequate: postconceptional initiation of supplement use or no use at all.

^* P-value <0.05.

GA, gestational age.

Serum folate and embryonic development

Table 3 shows the associations between maternal periconception serum folate levels and Carnegie stages. A higher serum folate level was associated with accelerated embryonic development (β 0.21 [95% CI: 0.09–0.32], per IQR (14.6 nmol/l) increase in serum folate). Moreover, an insufficient serum folate level in the periconception period was associated with a 1.01-day delay in reaching the final Carnegie stage, compared to a sufficient serum folate level (Fig. 3). After excluding participants with a serum folate level above the upper LOD from all analysis, the effect estimates remained comparable (Supplementary Table S2).

Table 3.

Association between periconceptional lifestyle risk factors and embryonic development (Carnegie stages) and growth (crown-rump length).

	Carnegie stage (beta (95% CI)) N = 808		Crown-rump length (beta (95% CI)) N = 828
	Model 1a	Model 2b	Model 1a	Model 2b
Serum folate (per IQR increase = 14.6 nmol/l)	0.21 (0.09–0.32)*	0.19 (0.06–0.32)*	0.48 (0.17–0.80)*	0.51 (0.15–0.86)*
Serum folate
Sufficientc (≥25.5 nmol/l)	Reference	Reference	Reference	Reference
Insufficient (<25.5 nmol/l)	−0.36 (−0.56 to −0.16)*	−0.33 (−0.55 to −0.11)*	−0.89 (−1.45 to −0.33)*	−0.93 (−1.55 to −0.32)*

^aModel 1: GA.

^bModel 2: GA, maternal age at conception, BMI, ethnicity, educational level, parity, smoking and fetal sex consumption, physical activity, and folic acid supplement use (unless exposure of interest).

^cSufficient levels ≥25.5 nmol/l serum folate, based on recommendations for women of reproductive age.

^* P-value <0.05.

IQR, interquartile range; GA, gestational age.

Figure 3.

Effect plots of differences in embryonic outcomes. The association of sufficient folate levels is shown for embryonic development (A) and crown-rump length (B). The reported delay (in A) refers to the time taken to reach the final Carnegie stage (Stage 23) between groups with sufficient and insufficient folate levels. The reported percentages (in B) represent the differences in estimated CRL between groups with sufficient and insufficient folate levels. Insufficient plasma folate levels are defined as below the recommended threshold for women of reproductive age (25.5 nmol/l). CRL, crown-rump length.

Lifestyle risk factors, serum folate levels, and CRL

Linear mixed models showed that inadequate folic acid supplement use was associated with a smaller CRL (−0.46 mm [95% CI: −0.87 to −0.05]) compared with adequate folic acid supplement use (Table 2). Essentially, this indicates a reduction in CRL of −0.72 mm (−1.6%) at 10 + 2 weeks of gestation. Additionally, a higher serum folate level was associated with larger CRL (0.48 mm [95% CI: 0.17–0.80], per IQR (= 14.6 nmol/l) increase in serum folate) and a periconception insufficient serum folate level was associated with a reduction in CRL of 1.24 mm (−3.7%) at 10 + 2 weeks, compared with a sufficient serum folate level (Fig. 3). No associations were observed between periconception smoking, alcohol consumption, caffeine consumption, and embryonic growth. Additionally, in Model 1, a non-active lifestyle, compared to a very active lifestyle, was positively associated with CRL (β  =  1.53 mm [95% CI: 0.11–2.95]). After adjustment for confounders, particularly maternal age at conception, this association was no longer statistically significant.

Discussion

Main findings

This study investigated the association between periconceptional lifestyle risk factors and embryonic morphological development, as assessed by the Carnegie stages, in a large population-based periconception cohort. We found that no or inadequate folic acid supplementation and insufficient serum folate levels were associated with delayed embryonic development and reduced embryonic growth. No associations were demonstrated between smoking, alcohol consumption, caffeine consumption, and physical activity and embryonic development and growth.

Interpretation of findings and comparison with the literature

Our findings on serum folate and embryonic development are in line with only one previously conducted study. This study was performed within a prospective cohort study, including 234 high-risk pregnancies, and reported that low RBC folate concentrations were associated with delayed embryonic development in IVF/ICSI pregnancies (Parisi et al., 2017). RBC folate levels indicate long-term folate status, whereas serum folate is considered an indicator of more recent folate intake(World Health Organization, 2015). Although serum folate levels can fluctuate based on recent dietary intake, RBC folate measurement is more complex and costly. Therefore, it is relevant that our study observed the same association with serum folate, a more commonly used marker in clinical practice and large epidemiological studies. This makes serum folate a practical and accessible measure for assessing folate-related risks in the periconception period. No prior studies have investigated the association between folic acid supplement use and embryonic development. However, our findings do align with the majority of the studies on folic acid supplement use and growth assessed by CRL, showing a positive association with adequate periconception supplement use (Mook-Kanamori et al., 2010; van Uitert et al., 2013; Van Dijk et al., 2018). Several possible explanations exist for the effect of folic acid supplement use and serum folate on embryonic development and growth rate. First, folate plays a key role in one-carbon metabolism, providing essential one-carbon groups for critical processes such as DNA synthesis, protein synthesis, and epigenetic programming (Crider et al., 2012; Steegers-Theunissen et al., 2013). These processes are vital during early pregnancy, a period characterized by rapid cell division, growth, and high sensitivity to external influences and deficiencies may impair epigenetic programming, potentially leading to long-term health consequences for the child. Another potential mechanism involves the mechanistic target of rapamycin (mTOR), a folate sensor in the trophoblast cells important for placental amino acid transport and mitochondrial function (Rosario et al., 2016, 2017b; Silva et al., 2017). Folate deficiency has been shown to inhibit mTOR signaling, resulting in downregulation of placental amino acid transporters and impaired trophoblast function, which can restrict embryonic and fetal development (Yung et al., 2008; Roos et al., 2009; Chen et al., 2015; Rosario et al., 2017a).

In the present study, which included only spontaneous pregnancies, no association was found between smoking and embryonic development. In contrast, a previous prospective cohort study of 689 women in a high-risk population reported an inverse association between smoking and embryonic development (Pietersma et al., 2022). Notably, nearly half (49.2%) of that population consisted of IVF/ICSI pregnancies, and after stratifying by mode of conception, the effect of smoking was observed only in the IVF/ICSI group. This may be due to the precise dating of IVF/ICSI pregnancies, which allows for more accurate detection of small effects. Additionally, IVF/ICSI pregnancies may be more susceptible to external exposures due to the techniques involved in assisted reproduction (Klonoff-Cohen, 2005; Berntsen et al., 2019).

Alcohol and caffeine consumption and physical activity were not significantly associated with embryonic development in the present study. No prior studies investigated these associations, but studies that focused on CRL reported inconsistent results for alcohol and caffeine consumption (Mills et al., 1993; Bakker et al., 2010; Mook-Kanamori et al., 2010; van Uitert et al., 2013; Van Dijk et al., 2018). Regarding caffeine, one large cohort study among 1310 healthy pregnancies demonstrated an association between caffeine consumption of more than 540 mg/day (>6 units) and smaller CRL (Bakker et al., 2010). In the present study, the maximum reported caffeine consumption was 4 units per day, which may explain why we did not observe any significant associations. Regarding periconception alcohol consumption, only one small prospective cohort study (n = 77) in a high-risk population showed a negative association with CRL (van Uitert et al., 2013), whereas other studies found no association (Van Dijk et al., 2018; van der Windt et al., 2024).

Strengths and limitations

A key strength of this study is its large, population-based design, with many spontaneous pregnancies included in the periconception period. Additionally, the availability of serial 3D ultrasound data during the embryonic period is a unique feature as well. The 3D VR approach allowed for detailed observation of the embryonic morphology and an accurate and reliable CRL measurement. A potential limitation of this study is that pregnancy dating was based on the LMP, which is not the preferred method in clinical practice and can be less accurate due to variability in cycle length or inaccurate recall of the LMP (Savitz et al., 2002; Mahendru et al., 2012). However, to reduce the potential for gestational age misclassification, we only included pregnancies with regular cycles and a known LMP. Most of the lifestyle risk factors were explored using self-reported questionnaires, introducing the possibility of social desirability and recall bias and thus the chance of underreporting. This could result in an underestimation of the true effects of these risk factors. The included study population was relatively healthy, characterized by low exposure to adverse lifestyle factors, high adherence to periconceptional folic acid supplementation, and a high socioeconomic status. This may limit the generalizability of the findings to higher-risk group and additionally could make it more challenging to detect the effects of certain lifestyle risk factors. Finally, it is worth noting that the Carnegie staging system is based on ex vivo examination of external and internal characteristics. In vivo, however, not all these characteristics can be observed, which may limit the precision of embryonic staging.

Clinical and research implications

The findings of this large population-based cohort study suggest that health and lifestyle factors in the periconception period, particularly folic acid supplementation and serum folate levels, influence embryonic health. This underscores the critical importance of this period as a window for intervention. Although preconception counseling on folic acid supplementation is recommended in current guidelines, timely initiation before conception remains suboptimal in many populations. Our results highlight the importance of not only recommending folic acid use, but also ensuring early initiation and adequate folate levels prior to conception. Population-level strategies aimed at improving access and uptake, such as the provision of free folic acid supplements as implemented in some regions (e.g. Scotland), may support more effective preconception care. Future research should focus on identifying and evaluating strategies to improve timely uptake and adherence to folic acid supplementation before conception and assess whether these approaches translate into improved embryonic development and downstream pregnancy outcomes. Further research should prioritize using the Carnegie staging system in addition to crown-rump length, as it offers a more comprehensive marker of embryonic health. Furthermore, the incorporation of objective measures, such as biomarkers for alcohol (e.g. phosphatidylethanol) and smoking (e.g. cotinine or carbon monoxide testing) (Shipton et al., 2009; Wassenaar and Koch, 2015), alongside activity trackers to accurately monitor physical activity, would help to further elucidate the relationship between lifestyle risk factors and embryonic health and beyond.

In conclusion, inadequate folic acid supplement use and insufficient serum folate levels are associated with delayed embryonic development and reduced growth rates in the early stages of pregnancy. These findings highlight the importance of preconception initiation of folic acid supplementation to support embryonic health, underscoring the critical role of the periconception period. Incorporating embryonic morphologic development alongside growth in future research can provide deeper insights into the pathophysiology of external exposures on embryonic health, helping to determine how and when to intervene to optimize embryonic health and improve birth and childhood outcomes.

Supplementary data

Supplementary data are available at Human Reproduction online.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.

Authors’ roles

NG: Study design, data collection, data analysis, and manuscript drafting; AGP and MR: Study design, manuscript drafting, and supervision; AHJK: Software development and critical revision of the manuscript; AGMGJM, RG, VWVJ, and EAPS: Critical revision of the manuscript. All authors read and approved the final manuscript.

Funding

The Generation R Next Study is financially supported by the Erasmus University Medical Centre Rotterdam, the Erasmus University Rotterdam, and the Netherlands Organization for Health Research and Development. R.G. received funding from the Netherlands Organization for Health Research and Development (NWO, ZonMw VIDI 09150172110034, and NWO, ZonMW, grant number 05430052110007) and an European Research Council Starting Grant (ERC-2024-STG 101161004). V.W.V.J. received a grant from the Netherlands Organization for Health Research and Development (NWO, ZonMw 05430052110007) and a European Research Council Consolidator Grant (ERC-2014-CoG-648916).

Conflict of interest

The authors declare no conflicts of interest.