Childhood BMI after ART with frozen embryo transfer

Abstract

STUDY QUESTION

Does BMI at 7–10 years of age differ in children conceived after frozen embryo transfer (FET) compared to children conceived after fresh embryo transfer (fresh-ET) or natural conception (NC)?

SUMMARY ANSWER

BMI in childhood does not differ between children conceived after FET compared to children conceived after fresh-ET or NC.

WHAT IS KNOWN ALREADY

High childhood BMI is strongly associated with obesity and cardiometabolic disease and mortality in adulthood. Children conceived after FET have a higher risk of being born large for gestational age (LGA) than children conceived after NC. It is well-documented that being born LGA is associated with an increased risk of obesity in childhood, and it has been hypothesized that ART induces epigenetic variations around fertilization, implantation, and early embryonic stages, which influence fetal size at birth as well as BMI and health later in life.

STUDY DESIGN, SIZE, DURATION

The study ‘Health in Childhood following Assisted Reproductive Technology’ (HiCART) is a large retrospective cohort study with 606 singletons aged 7–10 years divided into three groups according to mode of conception: FET (n = 200), fresh-ET (n = 203), and NC (n = 203). All children were born in Eastern Denmark from 2009 to 2013 and the study was conducted from January 2019 to September 2021.

PARTICIPANTS/MATERIALS, SETTING, METHODS

We anticipated that the participation rate would differ between the three study groups owing to variation in the motivation to engage. To reach the goal of 200 children in each group, we invited 478 in the FET-group, 661 in the fresh-ET-group, and 1175 in the NC-group. The children underwent clinical examinations including anthropometric measurements, whole-body dual-energy x-ray absorptiometry-scan, and pubertal staging. Standard deviation scores (SDS) were calculated for all anthropometric measurements using Danish reference values. Parents completed a questionnaire regarding the pregnancy and the current health of the child and themselves. Maternal, obstetric, and neonatal data were obtained from the Danish IVF Registry and Danish Medical Birth Registry.

MAIN RESULTS AND THE ROLE OF CHANCE

As expected, children conceived after FET had a significantly higher birthweight (SDS) compared to both children born after fresh-ET (mean difference 0.42, 95% CI (0.21; 0.62)) and NC (mean difference 0.35, 95% CI (0.14; 0.57)). At follow-up (7–10 years), no differences were found in BMI (SDS) comparing FET to fresh-ET, FET to NC, and fresh-ET to NC. Similar results were also found regarding the secondary outcomes weight (SDS), height (SDS), sitting height, waist circumference, hip circumference, fat, and fat percentage. In the multivariate linear regression analyses, the effect of mode of conception remained non-significant after adjusting for multiple confounders. When stratified on sex, weight (SDS), and height (SDS) were significantly higher for girls born after FET compared to girls born after NC. Further, FET-girls also had significantly higher waist, hip, and fat measurements compared to girls born after fresh-ET. However, for the boys the differences remained insignificant after confounder adjustment.

LIMITATIONS, REASONS FOR CAUTION

The sample size was decided in order to detect a difference of 0.3 SDS in childhood BMI (which corresponds to an adult cardiovascular mortality hazard ratio of 1.034). Thus, smaller differences in BMI SDS may be overlooked. As the overall participation rate was 26% (FET: 41%, fresh-ET: 31%, NC: 18%), selection bias cannot be excluded. Regarding the three study groups, many possible confounders have been included but there might be a small risk of selection bias as information regarding cause of infertility is not available in this study.

WIDER IMPLICATIONS OF THE FINDINGS

The increased birthweight in children conceived after FET did not translate into differences in BMI, however, for the girls born after FET, we observed increased height (SDS) and weight (SDS) compared to the girls born after NC, while for the boys the results remained insignificant after confounder adjustment. Since body composition in childhood is a strong biomarker of cardiometabolic disease later in life, longitudinal studies of girls and boys born after FET are needed.

STUDY FUNDING/COMPETING INTEREST(S)

The study was funded by the Novo Nordisk Foundation (grant number: NNF18OC0034092, NFF19OC0054340) and Rigshospitalets Research Foundation. There were no competing interests.

TRIAL REGISTRATION NUMBER

ClinicalTrials.gov identifier: NCT03719703.

ART, frozen embryo transfer, BMI, anthropometrics, childhood growth

Introduction

High BMI in childhood is closely associated with increased risk of cardiovascular disease (Baker et al., 2007), diabetes (Twig et al., 2016a), and all-cause mortality in adult life (Twig et al., 2016b). Therefore, factors influencing childhood obesity should be of utmost interest to the individual and society. ART, notably with frozen embryo transfer (FET) is hypothesized to be related to childhood BMI, but this is not yet known.

Owing to improvements of the techniques and protocols applied in ART, an increasing number of infertile couples become parents after fertility treatment (CDC, 2019; Wyns et al., 2022). ART is associated with an increased risk of adverse obstetric and perinatal outcomes (Qin et al., 2016; Elias et al., 2020), however, the risk profiles differ between ART with FET and fresh embryo transfer (fresh-ET). Conception with FET increases the risk of high birthweight (macrosomia) and being born large for gestational age (LGA) (Pinborg et al., 2014; Spijkers et al., 2017; Berntsen and Pinborg, 2018; Terho et al., 2021a; Westvik-Johari et al., 2021), while fresh-ET increases the risk of preterm birth (PTB), low birthweight (BW), and being born small for gestational age (SGA) (Qin et al., 2016; Elias et al., 2020; Westvik-Johari et al., 2021). The use of FET has increased steadily during the last decades and currently accounts for more than 30% of ART cycles in Europe (Wyns et al., 2022), and more than 50% in USA and Australia (de Geyter et al., 2020).

Multiple studies have shown that both LGA and SGA increases the risk of obesity and development of the metabolic syndrome (obesity, hypertension, insulin resistance, and dyslipidaemia) later in life (Boney et al., 2005; Wang et al., 2007; Meas et al., 2010; Taal et al., 2013; Zhang et al., 2013; Johnsson et al., 2015; Hong and Chung, 2018; Kapral et al., 2018; Bizerea-Moga et al., 2022) and therefore long-term consequences of being conceived by ART are of special concern.

Most studies on growth in childhood in ART progeny have not been able to distinguish between FET and fresh-ET (Bay et al., 2014; Halliday et al., 2019; Ebdrup et al., 2021). The few studies that compare anthropometric measurements in children born after FET versus fresh-ET only include children in early childhood (primarily 0–5 years) focusing mainly on measurements between 0 and 2 years (Wennerholm et al., 1998; Hann et al., 2018; Ainsworth et al., 2019; Turner et al., 2020; Terho et al., 2021b). Only one small study has explored the anthropometric markers of child growth in older children conceived by both FET and fresh-ET (3.5–11 years) (Green et al., 2013). The results of most of the abovementioned studies show no differences in growth between children born after FET and fresh-ET, while two studies show contradictory findings regarding height (Green et al., 2013; Hann et al., 2018). Green et al. (2013) finds girls born after fresh-ET to be taller compared to FET and natural conception (NC), while Hann et al. (2018) showed children born after FET to be taller than children born after NC.

The aim of this study was to compare anthropometric measurements of children born after FET compared to fresh-ET and NC at 7–10 years of age with adjustment for relevant confounders. The hypothesis was that the increased BW caused by FET would increase childhood BMI.

Materials and methods

Study design and study population

This study is part of a larger project entitled: ‘Health in Childhood following Assisted Reproductive Technology’ (HiCART). This cohort study is the first study published on the entire HiCART cohort consisting of 606 singletons (292 boys, 314 girls) aged 7–10 years, born from November 2009 to December 2013 in the Eastern part of Denmark. The children were identified through the Danish IVF Registry and the Danish Medical Birth Registry and were recruited according to mode of conception:

• Singletons conceived by ART with FET (n = 200)

• Singletons conceived by ART with fresh-ET (n = 203)

• Singletons conceived naturally (n = 203)

A flowchart of the children included in the study is depicted in Fig. 1. Both children born after conventional IVF and ICSI were included in this study.

Figure 1.

Flowchart for inclusion of participants in the HiCART cohort. HiCART, Health in Childhood following Assisted Reproductive Technology.

We anticipated the highest participation rate in the FET-group and the lowest in the NC-group owing to variation in the motivation to engage. Therefore, for each child in the FET-group, we identified two children born after fresh-ET and four children born after NC, matched on year of birth and child sex. To reach the goal of 200 children in each group, we ended up inviting 478 in the FET-group, 661 in the fresh-ET-group and 1175 in the NC-group. Children born of mothers with diabetes mellitus or gestational diabetes, twins, siblings, children born after oocyte donation, and children with severe illnesses (assessed by specialized paediatrician) were excluded. Mothers of eligible children were invited to participate by mail (both letter and electronic mail) twice, then contacted by phone if they had not responded to the invitation. Invitations were sent until the desired number of participants was reached according to the sample size calculations.

Clinical examinations

All children underwent clinical examinations at the Department of Growth and Reproduction at Rigshospitalet, Copenhagen including anthropometric measurements (standing height, sitting height, waist- and hip circumference, and weight), whole-body dual-energy x-ray absorptiometry-scan (DXA), and pubertal staging. The investigators were blinded for mode of conception during the entire examination.

Standing height was measured without shoes to the nearest 0.1 cm with a wall-mounted Harpenden stadiometer (Holtain Ltd., Stadiometer, Crymych, UK). Sitting height was measured to the nearest 0.1 cm with Harpenden sitting height table (Holtain Ltd., Stadiometer, Crymych, UK). Waist- and hip circumference were measured manually to the nearest 0.1 cm at the umbilical level and at the level of the widest hip perimeter, respectively. Weight was measured in light clothing with a calibrated digital weight (Kern & Sohn GmbH, MPE 250K100PM, Balingen, Germany) to the nearest 0.1 kg.

All anthropometrics were measured in triplicate and mean values were used.

Whole-body DXA-scan was performed with light clothing to determine body composition (Lunar Progidy, GE Healthcare, Madison, WI, USA). Data are analyzed using enCORE software, version 16 mode ‘Enhanced’ (GE Healthcare, Madison, WI, USA). Calibration of the machine was performed using a quality assurance block daily and a spine phantom on a weekly basis, as recommended by the manufacturer. The fat percentage applied in this paper is ‘tissue fat %’ defined as body fat (g) divided by the difference between total weight and bone mineral content.

Pubertal staging was performed by a clinician according to Marshall and Tanner (1969, 1970). For girls, puberty was defined as breast stage ≥ 2, and for boys as testicular volume > 3 ml (measured by a Prader orchidometer to the nearest ml).

BMI was calculated for both children and parents as weight (kg) divided by height squared (m²). The height, weight, and BMI for the children are reported in raw numbers but were also converted into standard deviation scores (SDS) using the Lambda–Mu–Sigma method with a Danish child population as reference (Tinggaard et al., 2014). The same was done for BW (Marsal et al., 1996). This was done in order to allow comparison of children of different age and sex. LGA and SGA were defined as ≥ +2 SD and ≤ −2 SD, respectively.

Background information regarding mode of conception, culture duration, freezing method, endometrial preparation, pregnancy, and delivery were collected from the Danish Medical Birth Registry and the Danish IVF Registry. Both parents were invited to complete a questionnaire regarding education, ethnicity, and former and current health of themselves and their child. The response rate for the questionaries was 92% for the mothers and 78% for the fathers.

Statistical analysis

All statistical analyses were performed with R for Windows (version 3.6.1) and the Rstudio platform (version 1.2.5001; R Core Team 2020, R Foundation for Statistical Computing, Vienna, Austria). Siblings were excluded to ensure independency within the groups for statistical analysis.

Data are presented as mean (SD) and number (%) as appropriate.

Background characteristics in the three groups were compared pairwise (FET versus fresh-ET, FET versus NC, and fresh-ET versus NC). For continuous variables, univariate linear regression analysis was used, and results are reported as mean differences. For categorical variables a two-sided Chi-square test was used, and results are reported as risk differences. The same procedure was applied for the anthropometric outcomes where the results are shown for the whole cohort and stratified by sex.

All outcomes were adjusted for possible confounders using multiple linear regression analyses.

In four outcomes (BMI (SDS), weight (SDS), height (SDS), and fat percentage), we made stepwise adjustments and presented the results as forest plots. In ‘Model 0’, linear regression analysis was performed for the four outcomes and study group. In ‘Model 1’, sex, age at examination, and puberty were added to ‘Model 0’. Although participants were matched by age and sex, these variables were included in the adjustments to account for residual confounding. In ‘Model 2’, mother’s parity and BMI in early pregnancy were added to ‘Model 1’. In ‘Model 3’, maternal smoking during pregnancy and current maternal educational level were added to ‘Model 2’. In ‘Model 4’, breastfeeding was added to ‘Model 3’. In ‘Model 5’, BW (SDS) was added to ‘Model 4’.

The association between BW (SDS) and childhood BMI (SDS) within the three groups and between the two sexes was explored using linear regression analysis.

In the non-participant analysis, invited mothers from the three study groups were divided into those willing to participate and those either not responding or declining to participate. The participants versus the non-participants were compared by calculating mean differences (continuous variables using univariate linear regression model) and risk differences (categorical variables using a two-sided Chi-square test) between the three groups.

Sample size calculations

BMI (SDS) was the primary endpoint as BMI in childhood and adolescence show dose-dependent association with adult mortality, with an increase of 1 SDS at age 8 years showing a hazard ratio (HR) of 1.13 for boys and 1.06 for girls of a fatal cardiovascular event in adulthood (Baker et al., 2007). The minimal clinically important difference in BMI (SDS) was set as 0.3 SDS, corresponding to an increase in HR of 3.4% for boys. Thus, with equal size groups, a significance level of 5% and power of 90% there should be 191 children in each group to detect the minimal relevant difference (https://www.sealedenvelope.com/power/continuous-noninferior/, n.d.). With 200 children in each group, we anticipated that the sample size would be sufficiently large to compensate for degrees of freedom lost owing to adjustment for confounding factors.

Ethics approval

In accordance with the Declaration of Helsinki, all parents with full custody of the child gave written informed consent after receiving both written and oral information about the project. The study protocol was approved by the Ethics Committee of the Capital Region of Denmark (H-18000122) and The Danish Data Protection Agency (VD-2018-310).

Results

Participants

In total, 2314 mothers and their children were invited to participate in this study. Of these, 606 children were included (Fig. 1). The overall participation rate was 26% but differed between the three study groups (FET: 42%, fresh-ET: 31%, NC: 18%). According to the exclusion criteria, 36 mothers were not included although they responded positively to the invitation.

Background characteristics of the included study population are shown in Table 1. Children born after FET had – as expected – a significantly higher BW (SDS) compared to both children born after fresh-ET (mean difference 0.42, 95% CI (0.21; 0.62)) and NC (mean difference 0.35, 95% CI (0.14; 0.57)) (Table 1). The age of both parents at birth was higher for children born after ART compared to the parents of children born after NC. Mothers of children born after FET and NC had higher parity compared to the mothers of children born after fresh-ET and significantly fewer mothers smoked during pregnancy in the ART-groups compared to the NC-group. No differences were observed between the three groups with regards to gestational age, mother’s BMI at pregnancy, mother’s and father’s BMI at examination, self-estimated health, educational level, and ethnicity. Further, the proportion of children being SGA and LGA did not differ between the groups (Table 1).

Table 1.

Basic characteristics of the included children, their mothers, and their fathers in a study of childhood BMI.

	Frozen embryo transfer (FET)		Fresh embryo transfer (fresh-ET)		Naturally conceived (NC)		FET vs. fresh-ETa(ref: fresh-ET)	FET vs. NCa(ref: NC)	Fresh-ET vs. NCa(ref: NC)
	n		n		n
Child characteristics
Sex, (n (%))b	200		203		203
Girls		112 (56.0)		101 (49.8)		101 (49.8)	0.06 (−0.04; 0.16)	0.06 (−0.04; 0.16)	0.00 (−0.10; 0.10)
Gestational age, daysb	200	278 (12.9)	203	279 (11.0)	199	279 (10.9)	−0.99 (−3.34; 1.35)	−0.56 (−2.90; 1.79)	0.44 (−1.70; 2.58)
Gestational age (n (%))b	200		203		199
Premature (<37 weeks)		13 (6.5)		8 (3.9)		6 (3.0)	0.03 (−0.02; 0.07)	0.03 (−0.01; 0.08)	0.01 (−0.03; 0.05)
Mature (37–42 weeks)		177 (8.5)		191 (4.1)		184 (2.5)	−0.06 (−0.16; 0.00)	−0.04 (−0.10; 0.02)	0.02 (−0.04; 0.07)
Postmature (>42 weeks)		10 (5.0)		4 (2.0)		9 (4.5)	0.03 (−0.01; 0.07)	0.00 (−0.04; 0.05)	−0.03 (−0.07; 0.01)
Birthweight, gb	200	3584 (533)	203	3452 (483)	200	3471 (498)	131.91 (32.3; 231.5)	112.83 (11.4; 214.2)	−19.09 (−115.1; 76.9)
Birthweight, SDSb	200	0.20 (1.09)	203	−0.22 (1.00)		−0.16 (1.09)	0.42 (0.21; 0.62)	0.35 (0.14; 0.57)	−0.06 (−0.27; 0.14)
Birthweight for gestational age, SDS (n (%))	200		203		197
SGA (< −2SD)		5 (2.5%)		9 (4.4%)		7 (3.6%)	−0.02 (−0.06; 0.02)	−0.01 (−0.05; 0.03)	0.01 (−0.03; 0.05)
AGA		186 (93)		191 (94%)		186 (94%)	−0.01 (−0.06; 0.04)	−0.01 (−0.07; 0.04)	−0.00 (−0.05; 0.05)
LGA (>2SD)		9 (4.5%)		3 (1.5%)		4 (2.0%)	0.03 (−0.01; 0.07)	0.02 (−0.02; 0.06)	−0.01 (−0.04; 0.03)
Maternal characteristics
Age at delivery, yearsb	186	34.8 (4.2)	193	34.6 (4.5)	184	32.1 (5.0)	0.20 (−0.68; 1.08)	2.75 (1.80; 3.70)	2.55 (1.58; 3.52)
Fertilization (n (%))b	177		196		203
IVF		117 (66.1)		108 (55.1)		0 (0.0)	0.11 (0.01; 0.21)	–	–
ICSI		60 (33.9)		88 (44.9)		0 (0.0)	−0.11 (−0.21; −0.01)	–	–
Spontaneously		0 (0.0)		0 (0.0)		203 (100)	–	–	–
Parity (n (%))b	200		201		202
Nulliparous		108 (54.0)		160 (79.6)		94 (46.5)	−0.26 (−0.35; −0.16)	0.07 (−0.03; 0.18)	0.33 (0.24; 0.42)
BMI at 1. trimester of pregnancy, kg/m2 (n (%))b	199		196		193
<18.5		4 (2.0)		4 (2.0)		6 (3.1)	−0.00 (−0.03; 0.03)	−0.01 (−0.05; 0.03)	−0.01(−0.05; 0.03)
18.5–24.9		153 (76.9)		142 (72.5)		134 (69.4)	0.04 (−0.05; 0.14)	0.07 (−0.02; 0.17)	0.03 (−0.07; 0.13)
≥25.0		42 (21.1)		50 (25.5)		53 (27.5)	−0.04 (−0.13; 0.04)	−0.06 (−0.15; 0.03)	−0.02 (−0.11; 0.07)
Smoking during pregnancy (n (%))b	199		203		197
Yes		2 (1.0)		3 (1.5)		17 (9.5)	−0.00 (−0.03; 0.02)	−0.08 (−0.12; −0.03)	−0.07 (−0.12; −0.02)
Solely breastfeeding, month (n (%))c	169		169		163
0		10 (5.9)		14 (8.3)		5 (3.1)	−0.02 (−0.08; 0.04)	0.03 (−0.02; 0.08)	0.05 (−0.00; 0.11)
>0–4		53 (31.4)		51 (30.2)		73 (44.8)	0.01 (−0.09; 0.12)	−0.13 (−0.24; −0.02)	−0.15 (−0.26; −0.04)
>4		106 (62.7)		104 (61.5)		85 (52.1)	0.01 (−0.10; 0.12)	0.11 (−0.01; 0.22)	0.09 (−0.02; 0.21)
BMI at examination of child, kg/m2 (n (%))c	185		193		183
<18.5		1 (0.5)		2 (1.0)		2 (1.1)	−0.00 (−0.03; 0.02)	−0.01 (−0.03; 0.02)	−0.00 (−0.02; 0.02)
18.5–24.9		125 (67.6)		119 (61.7)		114 (62.3)	0.06 (−0.04; 0.16)	0.05 (−0.05; 0.16)	−0.01 (−0.11; 0.10)
≥25.0		59 (31.9)		72 (37.3)		67 (36.6)	−0.05 (−0.16; 0.05)	−0.05 (−0.15; 0.06)	0.01 (−0.10; 0.11)
Self-estimated health (n (%))c	183		191		184
Excellent/very good		123 (67.2)		124 (64.9)		134 (72.8)	0.02 (−0.08; 0.12)	−0.05 (−0.16; 0.04)	−0.07 (−0.18; 0.02)
Good		51 (27.9)		64 (33.5)		44 (23.9)	−0.06 (−0.16; 0.04)	0.04 (−0.06; 0.13)	0.10 (−0.00; 0.19)
Poor		9 (4.9)		3 (1.6)		6 (3.3)	0.03 (−0.01; 0.07)	0.02 (−0.03; 0.06)	−0.02 (−0.05; 0.02)
Educational level (n (%))c	183		191		184
Higher education >4 years		77 (42.1)		81 (42.4)		70 (38.0)	−0.00 (−0.11; 0.10)	0.04 (−0.07; 0.15)	0.04 (−0.06; 0.15)
Ethnicity other than Caucasian, n (%))c	183		191		184
Yes		30 (16.4)		20 (10.5)		23 (12.5)	0.06 (−0.02; 0.13)	0.04 (−0.03; 0.12)	−0.02 (−0.09; 0.05)
Paternal characteristics
Age at delivery, yearsc	156	37.6 (5.3)	152	37.0 (6.0)	157	33.9 (5.9)	0.64 (−0.63; 1.92)	3.77 (2.52; 5.02)	3.12 (1.79; 4.46)
BMI at examination of child, kg/m2 (n (%))c	156		156		160
<18.5		0 (0.0)		1 (0.6)		0 (0.0)	−0.01 (−0.03; 0.01)	–	0.01 (−0.01; 0.03)
18.5–24.9		71 (45.5)		67 (42.9)		69 (43.1)	0.03 (−0.09; 0.14)	0.02 (−0.09; 0.14)	−0.00 (−0.11; 0.11)
≥25.0		85 (54.5)		88 (56.4)		91 (56.9)	−0.02 (−0.14; 0.10)	−0.02 (−0.14; 0.09)	−0.00 (−0.12; 0.11)
Self-estimated health (n (%))c	156		156		160
Excellent/very good		102 (65.4)		98 (62.8)		90 (56.3)	0.03 (−0.09; 0.14)	0.09 (−0.02; 0.20)	0.07 (−0.05; 0.18)
Good		49 (31.4)		55 (35.3)		65 (40.6)	−0.04 (−0.15; 0.07)	−0.09 (−0.20; 0.02)	−0.05 (−0.17; 0.06)
Poor		5 (3.2)		3 (1.9)		5 (3.1)	0.01 (−0.03; 0.05)	0.00 (−0.04; 0.04)	−0.01 (−0.05; 0.03)
Educational level (n (%))c	156		156		160
Higher education >4 years		70 (44.9)		61 (39.1)		57 (35.6)	0.06 (−0.06; 0.17)	0.09 (−0.02; 0.21)	0.03 (−0.08; 0.15)
Ethnicity other than Caucasian (n (%))c	156		156		160
Yes		17 (10.9)		11 (7.1)		21 (13.1)	0.04 (−0.03; 0.11)	−0.02 (−0.10; 0.06)	−0.06 (−0.13; 0.01)

Values are shown as mean (SD).

a

Pairwise comparison of groups by calculating mean differences with 95% CI for continuous variables (using univariate linear regression model) and risk differences with 95% CI for categorical variables (using a two-sided Chi-square test).

b

Information from the Danish National IVF registry.

c

Information from questionnaire.

AGA, appropriate for gestational age; LGA, large for gestational age; SDS, standard deviation score; SGA, small for gestational age.

Anthropometrics and body composition

BMI and BMI (SDS) did not differ significantly between singletons born after FET, fresh-ET, or NC (mean difference BMI (SDS): FET versus fresh-ET 0.11, 95% CI (−0.11; 0.34), FET versus NC −0.04, 95% CI (−0.25; 0.17), fresh-ET versus NC −0.15, 95% CI (−0.36; 0.05)). Similar results were found regarding the secondary outcomes weight, weight (SDS), height, height (SDS), sitting height, waist circumference, hip circumference, fat, and fat percentage before adjusting for relevant confounders.

After adjustment for relevant confounders (child sex, age at examination, puberty, mother’s parity, BMI in early pregnancy and educational level, smoking in pregnancy, breastfeeding, and BW (SDS)), the differences between the three groups remained insignificant (Table 2). The step-by-step adjustments for BMI (SDS), weight (SDS), height (SDS), and fat percentage are shown as forest plots for the whole cohort in Fig. 2.

Figure 2.

Forest plots for multiple linear regression models: step-by-step adjustments are shown for the whole cohort. (A) BMI (SDS), (B) Weight (SDS), (C) Height (SDS), and (D) Fat percentage. Model 0: Outcome variable versus study group. Model 1: Model 0 + adjustment for child sex, age at examination, and puberty. Model 2: Model 1 + adjustment for mother’s parity and BMI in early pregnancy. Model 3: Model 2 + adjustment for maternal educational level and smoking in pregnancy. Model 4: Model 3 + adjustment for breastfeeding. Model 5: Model 4 + adjustment for birth weight (SDS). FET, frozen embryo transfer; fresh-ET, fresh embryo transfer; NC, natural conception.

Table 2.

Anthropometric results at 7–10 years for the included children.

	Frozen embryo transfer (FET)	Fresh embryo transfer (fresh-ET)	Naturally conceived (NC)	FET vs. fresh-ETaadjusted	FET vs. NCa adjusted	Fresh-ET vs. NCaadjusted
	n = 200	n = 203	n = 203
Age at examination, years	8.44 (0.56)	8.55 (0.51)	8.58 (0.58)	−0.11 (−0.24; 0.02)	−0.17 (−0.30; −0.04)	−0.06 (−0.19; 0.07)
Weight, kg	30.51 (5.67)	30.28 (5.67)	30.51 (5.28)	0.75 (−0.36; 1.86)	0.63 (−0.48; 1.73)	−0.12 (−1.25; 1.00)
Weight, SDS	0.22 (1.07)	0.10 (1.06)	0.14 (1.00)	−0.00 (−0.24; 0.24)	0.13 (−0.11; 0.37)	0.13 (−0.11; 0.38)
Height, cm	134.78 (6.64)	135.06 (6.13)	134.71 (6.70)	0.26 (−1.03; 1.56)	0.61 (0.68; 1.89)	0.34 (−0.97; 1.65)
Height, SDS	0.25 (1.00)	0.17 (0.96)	0.07 (1.06)	0.03 (−0.20; 0.27)	0.23 (−0.01; 0.47)	0.20 (−0.04; 0.44)
BMI, kg/m2	16.70 (2.24)	16.49 (2.15)	16.72 (1.95)	−0.02 (−0.50; 0.46)	0.06 (−0.42; 0.53)	0.08 (−0.41; 0.56)
BMI, SDS	0.16 (1.16)	0.04 (1.10)	0.20 (1.01)	−0.04 (−0.29; 0.21)	−0.02 (−0.27; 0.23)	0.02 (−0.23; 0.27)
Sitting height, cm	71.47 (3.43)	71.58 (3.33)	71.50 (3.14)	0.00 (−0.66; 0.66)	0.10 (−0.55; 0.76)	0.10 (−0.57; 0.77)
Waist circumference, cm	59.08 (6.43)	58.51 (5.45)	58.75 (5.61)	0.98 (−0.22 ; 2.18)	0.70 (−0.49; 1.89)	−0.28 (−1.50; 0.93)
Hip circumference, cm	70.13 (6.06)	69.79 (6.75)	70.00 (6.03)	0.83 (−0.56; 2.21	0.80 (−0.58; 2.18)	−0.03 (−1.43; 1.38)
Total fat, kg	8.49 (3.34)	8.43 (3.48)	8.48 (3.17)	0.35 (−0.31; 1.14)	0.32 (−0.34; 0.98)	−0.03 (−0.70; 0.64)
Total fat percentage, %	28.16 (6.77)	28.08 (6.50)	28.20 (6.24)	0.31 (−0.98; 1.59)	0.34 (−0.93; 1.62)	0.04 (−1.26; 1.34)
Pubertyb, n (%)	15 (7.5)	18 (8.9)	24/203 (11.8)	–	–	–

Values are shown as mean (SD).

a

 Differences between groups are calculated as mean differences with 95% CI for continuous variables (using univariate linear regression model) and risk differences with 95% CI for categorical variables (using a two-sided Chi-square test). Mean differences are adjusted for confounders (sex, age at examination, puberty, parity, pregestational BMI, smoking during pregnancy, educational level (mother), breastfeeding, birth weight (SDS)) with multiple linear regression analysis.

b

 Puberty was performed according to ‘Marshall and Tanner’ and defined for girls as breast stage ≥2, and for boys as testicular volume >3 ml.

SDS, standard deviation score.

Outcomes after stratification on child sex

In the adjusted analyses, none of the anthropometric outcomes differed between the three groups for the boys (Table 3), but for the girls born after FET, we found significantly higher weight (SDS) and height (SDS) compared to girls born after NC (adj. mean differences, weight (SDS): 0.35, 95% CI (0.03; 0.67), height (SDS): 0.43 (0.11; 0.76)). Further, significantly higher waist circumference, hip circumference, total fat, and fat percentage were observed for girls in the FET-group compared to those in the fresh-ET-group (adj. mean differences, waist circumference: 1.75, 95% CI (0.06; 3.34), hip circumference: 2.83, 95% CI (1.07; 4.59), total fat: 1.02, 95% CI (0.12; 1.92), fat percentage 1.79, 95% CI (0.03; 3.54)) (Table 4).

Table 3.

Anthropometric results at 7–10 years for the included boys.

	Frozen embryo transfer (FET)	Fresh embryo transfer (fresh-ET)	Naturally conceived (NC)	FET vs. fresh-ETaadjusted	FET vs. NCaadjusted	Fresh-ET vs. NCaadjusted
	n = 88	n = 102	n = 102
Age at examination, years	8.46 (0.56)	8.54 (0.51)	8.50 (0.59)	−0.13 (−0.33; 0.06)	−0.10 (−0.30; 0.10)	0.03 (−0.17; 0.24)
Weight, kg	30.63 (5.94)	30.90 (5.86)	30.58 (5.19)	−0.16 (−1.84; 1.53)	0.22 (−1.51; 1.95)	0.38 (−1.34; 2.10)
Weight, SDS	0.12 (1.15)	0.13 (1.09)	0.13 (1.02)	−0.33 (−0.68; 0.02)	−0.14 (−0.50; 0.22)	0.19 (−0.17; 0.55)
Height, cm	135.40 (7.06)	136.17 (5.92)	134.73 (6.13)	−0.60 (−2.44 (1.25)	0.26 (−1.64; 2.16)	0.85 (−1.03; 2.74)
Height, SDS	0.18 (1.07)	0.24 (0.91)	0.02 (0.97)	−0.27 (−0.60; 0.06)	0.01 (−0.32; 0.34)	0.27 (−0.06; 0.61)
BMI, kg/m2	16.59 (2.15)	16.55 (2.17)	16.76 (1.99)	−0.56 (−1.28; 0.16)	−0.42 (−1.16; 0.32)	0.14 (−0.59; 0.88)
BMI, SDS	0.08 (1.22)	0.04 (1.18)	0.22 (1.04)	−0.31 (−0.70; 0.07)	−0.29 (−0.68; 0.10)	0.03 (−0.36; 0.42)
Sitting height, cm	71.64 (3.73)	72.13 (2.97)	71.63 (2.84)	−0.32 (−1.30; 0.65)	−0.20 (−1.20; 0.80)	0.12 (−0.87; 1.12)
Waist circumference, cm	59.63 (6.34)	59.23 (5.74)	59.61 (5.69)	0.18 (−1.47; 1.83)	0.10 (−1.60; 1.79)	−0.08 (−1.76; 1.60)
Hip circumference, cm	69.14 (5.94)	70.29 (7.42)	69.57 (6.26)	−1.48 (−3.66; 0.70)	−0.00 (−2.24; 2.24)	1.47 (−0.75; 3.70)
Total fat, kg	7.61 (3.38)	8.08 (3.67)	7.83 (3.14)	−0.37 (−1.35; 0.61)	−0.02 (−1.01; 1.00)	0.37 (−0.63; 1.37)
Total fat percentage, %	24.99 (6.65)	26.16 (6.58)	25.88 (6.05)	−1.33 (−3.23; 0.57)	−0.36 (−2.31; 1.60)	0.97 (−0.97; 2.91)
Pubertyb, n (%)	0/85 (0.0)	2/100 (2.0)	1/98 (1.0)	–	–	–

Values are shown in means (SD).

a

 Differences between groups are calculated as mean differences with 95% CI for continuous variables (using univariate linear regression model) and risk differences with 95% CI for categorical variables (using a two-sided Chi-square test). Mean differences are adjusted for confounders (sex, age at examination, puberty, parity, pregestational BMI, smoking during pregnancy, educational level (mother), breastfeeding, birth weight (SDS)) with multiple linear regression analysis.

b

 Puberty assessment was performed according to ‘Marshall and Tanner’ and defined as testicular volume >3 ml.

SDS, standard deviation score.

Table 4.

Anthropometric results at 7–10 years for the included girls.

	Frozen embryo transfer (FET)	Fresh embryo transfer (fresh-ET)	Naturally conceived (NC)	FET vs. fresh-ETaadjusted	FET vs. NCaadjusted	Fresh-ET vs. NCaadjusted
	n = 112	n = 101	n = 101
Age at examination, years	8.43 (0.56)	8.55 (0.50)	8.65 (0.57)	−0.11 (−0.30; 0.07)	−0.21 (−0.38; −0.05)	−0.10 (−0.30; 0.07)
Weight, kg	30.42 (5.43)	29.65 (5.48)	30.43 (5.39)	1.58 (0.11; 3.06)	1.03 (−0.40; 2.46)	−0.56 (−2.05; 0.94)
Weight, SDS	0.30 (1.01)	0.07 (1.03)	0.16 (0.99)	0.30 (−0.03; 0.63)	0.35 (0.03; 0.67)	0.05 (−0.28; 0.39)
Height, cm	134.30 (6.28)	133.95 (6.16)	134.68 (7.26)	1.04 (−0.80; 2.88)	0.99 (−0.79; 2.78)	−0.04 (−1.91; 1.83)
Height, SDS	0.30 (0.95)	0.11 (1.00)	0.12 (1.14)	0.31 (−0.03; 0.65)	0.43 (0.11; 0.76)	0.12 (−0.22; 0.47)
BMI, kg/m2	16.78 (2.31)	16.43 (2.13)	16.68 (1.93)	0.48 (−0.17; 1.13)	0.41 (−0.22; 1.04)	−0.07 (−0.72; 0.59)
BMI, SDS	0.23 (1.10)	0.05 (1.03)	0.18 (0.98)	0.21 (−0.13; 0.55)	0.17 (−0.15; 0.50)	−0.04 (−0.38; 0.31)
Sitting height, cm	71.34 (3.18)	71.02 (3.58)	71.36 (3.43)	0.29 (−0.62; 1.20)	0.36 (−0.52; 1.24)	0.07 (−0.85; 0.99)
Waist circumference, cm	58.64 (6.50)	57.79 (5.07)	57.87 (5.42)	1.75 (0.06; 3.44)	1.23 (−0.41; 2.87)	−0.52 (−2.37; 1.20)
Hip circumference, cm	70.91 (6.07)	69.29 (6.00)	70.43 (5.79)	2.83 (1.07; 4.59)	1.51 (−0.18; 3.22)	−1.31 (−3.10; 0.47)
Total fat, kg	9.18 (3.16)	8.78 (3.27)	9.13 (3.09)	1.02 (0.12; 1.92)	0.59 (−0.28; 1.46)	−0.43 (−1.34; 0.48)
Total fat percentage, %	30.65 (5.78)	30.02 (5.84)	30.54 (5.54)	1.79 (0.03; 3.54)	0.85 (−0.84; 2.55)	−0.93 (−2.71; 0.85)
Pubertyb, n (%)	15/111 (13.5)	16/101 (15.8)	23/101 (22.8)	–	–	–

Values are shown in means (SD).

a

 Differences between groups are calculated as mean differences with 95% CI for continuous variables (using univariate linear regression model) and risk differences with 95% CI for categorical variables (using a two-sided Chi-square test). Differences between groups are calculated as mean differences with 95% CI for continuous variables (using univariate linear regression model) and risk differences with 95% CI for categorical variables (using a two-sided Chi-square test). Mean differences are adjusted for confounders (sex, age at examination, puberty, parity, pregestational BMI, smoking during pregnancy, educational level (mother), breastfeeding, birth weight (SDS)) with multiple linear regression analysis.

b

 Puberty was performed according to ‘Marshall and Tanner’ and defined as breast stage ≥2.

SDS, standard deviation score.

Association between birthweight and childhood BMI

The association between BW (SDS) and childhood BMI (SDS) was studied and an overall positive effect size of 0.07, 95% CI (−0.01; 0.16) was found, in agreement with previous findings (Kapral et al., 2018). The association was also investigated for the individual three modes of conception and by sex. Results are shown in Supplementary Table S1. The association was stronger for the girls than the boys. The association was strongest in the fresh-ET-group compared to FET and NC.

Non-participants

The results of the non-participant analysis are shown in Supplementary Table S2. The mothers (n = 1672) who declined to participate or did not respond to the invitation were included in a non-participant analysis showing only a few differences between participants (p) and non-participants (np) within the three study groups (Supplementary Table S2). BW (SDS) was significantly higher in FET_p compared to FET_np (mean difference 0.27, 95% CI (0.08; 0.46)), which only amplifies the possible risks associated with BW in our study. Parity (expressed as parity >1) was higher (risk difference 0.11, 95% CI (0.01; 0.20)) and smoking during pregnancy marginally lower (risk difference −0.05, 95% CI (−0.09; −0.02)) in FET_p compared to FET_np. Regarding the fresh-ET-group, the mothers had higher gestational age (mean difference 4.49 days, 95% CI (1.95; 7.03)), and fewer preterm deliveries (risk difference −0.06, 95% CI (−0.11; −0.02)) compared to non-participants. BW (SDS) was comparable for participants and non-participant in the fresh-ET-group. The NC_p-mothers were slightly older at delivery compared to NC_np-mothers.

Discussion

Anthropometric measurements of 7- to 10-year-old singletons born after FET were compared to singletons born after fresh-ET and singletons born after NC in this large retrospective cohort study. As expected, mean BW was significantly higher in FET-singletons compared to singletons conceived after fresh-ET and NC. Despite this, we found no differences in BMI or other anthropometric measurements between the three groups of children at 7–10 years of age. Stratified on sex and adjusted for confounders, girls born after FET had significantly higher weight (SDS) and height (SDS) compared to girls born after NC. Additionally, girls born after FET had higher waist circumference, hip circumference, and body fat compared to girls born after fresh-ET.

To the best of our knowledge, this is the first large clinical study investigating anthropometrics of children born after ART with FET and fresh-ET at 7–10 years of age. Even though the clinical relevance of the increased BW in children born after FET is not known it has been shown that high BW is associated with higher BMI in childhood (Kapral et al., 2018; Bizerea-Moga et al., 2022). In this study, focus has been put on BMI during childhood since increased childhood BMI increases the risk of metabolic disease later in life (Baker et al., 2007; Twig et al., 2016a,b). Former studies addressing the BMI of children born after FET and fresh-ET were primarily based on registers or self-reported measurements (Hann et al., 2018; Magnus et al., 2021; Terho et al., 2021b), and focused primarily on children aged 0–5 years (Wennerholm et al., 1998; Ainsworth et al., 2019; Turner et al., 2020; Terho et al., 2021b). Only few children born after FET were included in the majority of the studies (Green et al., 2013; Ainsworth et al., 2019) and a higher BW in the FET-group compared to NC-children, which is the basis for our hypothesis, was not always present (Wennerholm et al., 1998; Ainsworth et al., 2019; Turner et al., 2020; Terho et al., 2021b).

The majority of existing studies reported no significant differences regarding BMI, height, or weight when comparing FET, fresh-ET, and NC (Wennerholm et al., 1998; Ainsworth et al., 2019; Terho et al., 2021b), which is in line with our current study. Other findings are somewhat contradictory to ours. One study showed that girls born after fresh-ET were taller than girls born after FET and NC (Green et al., 2013) whereas another study presented data showing that children born after FET were slightly taller than NC-children (Hann et al., 2018). However, the latter study had no data on parental height, and there was a concern about selection bias (Hann et al., 2018). Another study revealed that children born after FET had higher weight (SDS) than children born after fresh-ET at the age of 2 years, but no significant differences regarding weight, height, and BMI were observed at the age of 17 years (Magnus et al., 2021). In contrast, Turner et al. (2020) found a marginally increased weight (SDS) in 5-year-old children born after fresh-ET compared to NC. Thus, it could be speculated that the difference in weight at birth between the FET, fresh-ET, and NC may attenuate over time.

Multiple independent factors may influence childhood weight gain. In the adjusted regression analyses, we have taken as many confounders as possible into account. Mothers with diabetes before or during pregnancy were excluded because a strong correlation between maternal diabetes and childhood obesity is evident (Vlachová et al., 2015; Lowe et al., 2019). Maternal obesity (Ohlendorf et al., 2019; Voerman et al., 2019), low educational level of the parents (Chaparro and Koupil, 2014; Bramsved et al., 2018), and lack of breastfeeding (Yan et al., 2014; Mantzorou et al., 2022) are confounders that all increase the risk of childhood obesity. Further, being first-born (Bohn et al., 2021) and smoking during pregnancy (Di et al., 2022) results in lower mean BW.

The strengths of the current study are the age group in which in the children has been examined and the extensive information regarding possible confounders. Further, the substantial number of children included in the FET-group is a significant strength of this study. All children were seen in the same setting, and only a limited number of healthcare professionals were involved in the examinations, reducing interrater variability. In the HiCART study, we had few missing values from the clinical examinations. A high response rate was obtained from both mothers and fathers on the questionnaire, which provided valuable information regarding possible confounders. The access to additional and reliable information from the National Danish IVF Registry and Danish Medical Birth Registry using the unique personal identification number enabled the construction of a thorough non-participants analysis, showing only small differences between participants and non-participants. Hence the results of this study are less likely to be skewed by selection bias. Regarding the three study groups, many possible confounders have been described in Table 1 but there might be a small risk of selection bias since information regarding cause of infertility is not available in this study.

Studies have shown that hormone replacement therapy for endometrial preparation in FET-cycles (HRT-FET), also known as programmed FET or artificial cycle FET (AC-FET), affects the severity of the adverse obstetric and perinatal outcomes (Busnelli et al., 2022; Roelens et al., 2022). The proposed theory behind the differences is that the absence of a corpus luteum may lead to increased risk of pre-eclampsia and macrosomia (von Versen-Höynck et al., 2019a,b). In a HRT-FET-cycle (with artificial hormonal substitution and no corpus luteum), an increased risk of LGA, macrosomia, hypertensive disorders in pregnancy, PTB, post-term birth, post-partum hemorrhage, caesarean section, placental abruption, and placenta accreta are seen compared to a natural FET-cycle (without hormonal substitution and with a corpus luteum) (Asserhøj et al., 2021; Moreno-Sepulveda et al., 2021; Rosalik et al., 2021). In our study, only 6.8% of the children born after FET were born after a HRT-FET-cycle: in a sub-analysis comparing HRT-FET with natural FET, no significant differences were found regarding the anthropometric outcomes (data not shown). The studies on growth in childhood after FET by Terho et al. and Green et al. were the only two with information on cycle protocol. In these studies, only 10–20% of the FET-cycles were performed with hormonal substitution, thereby only having a minority of children in the ‘high-risk’ HRT-FET-group, supporting their findings of no/minimal difference in growth between the three groups of children (Green et al., 2013; Terho et al., 2021b).

In the adjusted analyses on anthropometric results stratified on sex, we observed significantly higher weight (SDS) and height (SDS) in girls born after FET compared to girls born after NC. Further, significant differences were seen for waist circumference, hip circumference, total fat, and fat percentage comparing girls born after FET to girls born after fresh-ET, although the latter seems more related to the decreased figures for the girls born after fresh-ET than higher values for the girls born after FET. For the boys born after FET, we did not observe any significant differences compared to either boys born after fresh-ET or NC. Other studies looking at parameters related to metabolism have seen an increased risk of a dysmetabolic phenotype in girls born by mothers with type 1 diabetes compared to boys, despite the absence of any apparent explanation (Lohse et al., 2018; Overgaard et al., 2022). We therefore speculate that the sex of the child might also play a role in the phenotype of children born after FET.

As the overall participation rate was 26% in the three groups selection bias cannot be excluded, especially in the NC-group with a participation rate of 18%. However, the non-participant analysis showed few differences, thereby minimizing the likelihood of selection bias. Another limitation of this study is the inclusion of children born after ART where older methods have been utilized. Today vitrification is the preferred freezing method and blastocyst transfer is primarily used. In the current study, slow-freeze was used in all the FET-treatments (9% missing data). In the FET and fresh-ET-groups combined, 76% received cleavage stage embryos and 5% blastocysts at transfer day (19% missing data). The distribution was the same in the two groups. Therefore, the present results do not fully represent current ART-treatments. However, studies have shown that both LGA and pre-eclampsia is present in FET both after vitrified/warmed blastocyst and slow freezing of cleavage stage embryo indicating that different freezing procedures cause similar adverse outcomes (Kaartinen et al., 2016; Ernstad et al., 2019).

Procedures involved in ART (e.g. hormonal substitution, culture medium, culture duration, and freezing/thawing procedures (Gaume et al., 2021)) may induce epigenetic variations around fertilization, implantation, and in the early embryonic stages, which may influence intrauterine growth and length and weight at birth (Jiang et al., 2017; Mani et al., 2020) although some argue that the changes of the epigenome resulting from ART are transient and do not persist into adolescence (Penova-Veselinovic et al., 2021). Cells from the placenta seem to be especially susceptible to epigenetic modification related to ART and infertility (Nelissen et al., 2013; Choufani et al., 2019). Some of these modifications, for example differences in the methylation of the glucocorticoid receptor gene, seems to have an influence on BW and the risk of being born LGA (Filiberto et al., 2011). Our results of altered anthropometric outcomes in girls from FET may support the theory that effects of possible ART-induced epigenetic modification during the early embryonic stages cause adverse health trajectories later in life.

Conclusion

Increased BW in children conceived after FET did not translate into differences in BMI, however, for the girls born after FET, we observed increased height and weight compared to the girls born after NC while no differences were seen for the boys. Since body composition in childhood is a strong biomarker of cardiometabolic disease later in life, longitudinal studies of girls and boys born after FET are needed to explore the long-term effects of embryo cryopreservation.

Data availability

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

Authors’ roles

T.D.C., E.R.H., G.G., P.L.M., R.B.J., and A.P. contributed to the study conception and design. L.L.A., I.M., G.F.H., and R.B.J. contributed to acquisition of data. L.L.A., I.M., R.B.J., and A.P. participated in the analysis of the data. L.L.A., I.M., T.D.C., E.R.H., G.G., K.M.M., P.L.M., R.B.J., and A.P. participated in the interpretation of the data. L.L.A. drafted the manuscript. All authors have critically reviewed the manuscript and approved the final version.

Funding

Novo Nordisk Foundation (NNF18OC0034092, NFF19OC0054340) and Rigshospitalets Research Foundation.

Conflict of interest

The authors have no conflict of interests to declare.