Impact of maternal fasting blood glucose in pregnancy on body composition, anthropometric, and metabolic outcomes in newborns and 3-Year-Old offspring: results from the lifestyle in pregnancy and offspring study

Abstract

Background

Gestational diabetes mellitus is becoming increasingly prevalent and has been associated with adverse outcomes for both mothers and their children. Prenatal exposure to maternal obesity and hyperglycemia has been associated with higher risks of macrosomia, neonatal hypoglycemia and later metabolic disorders. However, because diagnostic fasting glucose thresholds vary internationally, it remains unclear how different cut-offs, as well as fasting glucose as a continuous measure, relate to offspring growth and metabolism. This study investigated whether maternal fasting glucose predicts offspring body composition, anthropometry and metabolic outcomes at birth and at 3 years.

Methods

The analysis used data from the Lifestyle in Pregnancy (LiP) study and its 3-year follow-up, the LiPO study. Of 301 women in LiP, 157 participated in LiPO, and 75 children had DXA assessments. Maternal fasting plasma glucose at 28 weeks was analyzed continuously and across thresholds from 4.6 to 5.8 mmol/L using regression models adjusted for maternal age and pre-pregnancy BMI. Offspring outcomes included birth weight, birth weight z-score, BMI z-score, fat and fat-free mass, lipids and blood pressure.

Results

Higher maternal fasting glucose was associated with a higher birth weight z-score. At 3 years, no significant associations were found for BMI z-score or metabolic markers, although small, non-significant trends toward higher fat and fat-free mass were observed. Threshold analyses showed that glucose ≥ 5.6 mmol/L predicted higher birth weight and increased large-for-gestational-age risk, and glucose ≥ 5.3 mmol/L predicted higher cord c-peptide. No threshold was linked to metabolic differences at 3 years.

Conclusion

Maternal hyperglycemia was associated with adverse neonatal outcomes but not with offspring adiposity at 3 years. Longer-term follow-up, including the ongoing 15-year LiP data, is needed to determine whether mild maternal hyperglycemia has later metabolic consequences.

Trial registration

Both the LiP and LiPO studies received approval from the Regional Ethics Committee of Southern Denmark (S-20070058) and were registered with the Danish Data Protection Agency. The LiP study was registered at www.clinicaltrials.gov (NCT00530439) with a registration date of September 13, 2007. The LiPO follow-up, initiated in 2010 while the LiP study was still ongoing, was registered at www.clinicaltrials.gov (NCT01918319) to compare offspring of LiP participants.

Background

Gestational diabetes mellitus (GDM) represents a significant public health concern, with maternal obesity being one of its major risk factors. Approximately 17% of pregnant women with obesity develop GDM, compared to only 1–3% in the general population [1]. The rising prevalence of overweight and obesity among women of reproductive age, including during pregnancy, amplifies this concern [2]. Epidemiological data reveal considerable variation in obesity prevalence among pregnant women, with rates of about 33% in the United States, 20% in the United Kingdom, 15% in Australia, and 16% in Denmark [1–3]. Pregnant women with obesity exhibit increased insulin resistance, altered insulin responses, and elevated concentrations of inflammatory cytokines compared with women of normal weight, both before conception and throughout pregnancy [4]. These metabolic and inflammatory alterations contribute to a higher susceptibility to developing GDM. In turn, maternal obesity and GDM are associated with an increased risk of adverse pregnancy outcomes, including pre-eclampsia, cesarean delivery, and fetal macrosomia [5, 6].

The consequences of GDM and maternal hyperglycemia extend beyond maternal health, exerting a significant impact on offspring [7]. Prenatal exposure to maternal obesity and GDM has been associated with an increased risk of macrosomia and neonatal hypoglycemia, as evidenced by studies such as the Hyperglycemia and Adverse Pregnancy Outcome (HAPO) study [8].

Moreover, children born to mothers with GDM are predisposed to obesity and metabolic disorders later in life, thereby increasing their susceptibility to insulin resistance, type 2 diabetes, and cardiovascular complications [9, 10]. However, the long-term metabolic impacts of GDM on offspring health remain insufficiently documented, underscoring the need for further research to examine these associations and inform preventive strategies to mitigate their far-reaching health impacts [11]. Given the well-documented correlation, diagnostic classification, treatment, and outcomes depend on the cut-off values used for GDM during an oral glucose tolerance test (OGTT). Therefore, selecting and standardising these values is crucial for accurate identification and management across populations and healthcare settings. In Denmark, screening for GDM is primarily risk-factor based. Established risk factors comprise a history of GDM, previous delivery of a macrosomic infant (≥ 4,500 g), pre-pregnancy BMI ≥ 27 kg/m², a first-degree family history of diabetes, PCOS, and multiple gestation. Women with at least one of these risk factors are offered a diagnostic 2-hour 75 g oral glucose tolerance test (OGTT) at 24–28 weeks’ gestation. Those with a prior GDM diagnosis or at least two risk factors are additionally offered early testing at 10–20 weeks’ gestation. Furthermore, the presence of glucosuria at any time during pregnancy, or other clinical indicators such as an ultrasonographic fetal weight estimate > + 22% above the mean or polyhydramnios, also constitute indications for diagnostic testing [12]. A 2-hour glucose value of 9.0 mmol/L is used as the sole diagnostic criterion in Denmark [13]. Danish studies on GDM applying the WHO2013 criteria, such as the Lifestyle Intervention in Pregnancy (LiP) study [14], the Odense Child Cohort (OCC) [15], and the SugarMum study [13] suggest that a diagnostic fasting threshold of 5.1 mmol/L and 2-hour threshold of 8.5 mmol/L would classify approximately 20–40% of pregnant women in Denmark with GDM. Of these, approximately 90% of GDM diagnoses would be based on the fasting glucose [13–15].

Changing to the WHO2013 diagnostic criteria for GDM, with decreased glucose thresholds, has resulted in a significant increase in GDM prevalence in many countries [16]. In Denmark, register-based national cohort studies report an increase in GDM prevalence from 1.9% in 2004 to 4.4% in 2017 and 6% in 2023 [17, 18].

The discrepancy between thecurrent Danish and WHO2013 diagnostic criteria reveals a group of women who meet the WHO2013 criteria but do not meet the Danish criteria for GDM. This subgroup, currently untreated, may potentially face an elevated risk of both adverse perinatal and long-term outcomes for both mothers and offspring. Therefore, this study examined whether different diagnostic thresholds for GDM predict offspring body composition, anthropometrics and metabolic status at birth and at 3 years.

Methods

This secondary analysis examined the relationship between maternal fasting blood glucose levels measured at 28 weeks of gestation via a 75-g oral OGTT and offspring assessments at birth and at 3 years of age. The initial analyses evaluated maternal fasting plasma glucose (mmol/L) as a continuous exposure. Subsequently, we examined the same outcomes across different clinically relevant fasting glucose cut-off values according to potential diagnostic criteria for gestational diabetes mellitus (GDM), using thresholds of ≥ 5.1, ≥ 5.3, and ≥ 5.6 mmol/L. These thresholds were selected because they represent distinct diagnostic approaches used internationally. A fasting glucose cut-off of ≥ 5.1 mmol/L corresponds to the WHO2013 criteria. This threshold was derived from the HAPO study, which demonstrated a continuous association between increasing maternal fasting plasma glucose concentrations and adverse perinatal outcomes, including LGA, primary caesarean section, and preeclampsia [8]. Importantly, these associations were consistent across study centres, supporting their generalizability. Based on these observations, the International Association of Diabetes in Pregnancy Study Groups (IADPSG) recommended in 2010 the adoption of a single set of worldwide consensus diagnostic criteria for GDM, incorporating a fasting venous plasma glucose threshold of 5.1 mmol/L [19].

In contrast, a threshold of ≥ 5.3 mmol/L has been implemented in several European countries, including Norway and Finland [20, 21]. This approach is supported by evidence from recent studies, such as a Spanish cohort analysis, which demonstrated increased risks of LGA and gestational hypertension at fasting glucose levels above 5.3–5.5 mmol/L [22].

Finally, a threshold of ≥ 5.6 mmol/L may be regarded as most representative of the current Danish approach. A population-based analysis suggested that a locally adapted cut-off in the range of 5.5–5.7 mmol/L provides the greatest classification accuracy, particularly when LGA is considered alongside excess neonatal adiposity and neonatal hyperinsulinemia as composite outcomes of interest. Within the Danish context, such thresholds correspond to GDM prevalence rates ranging from 14.4% to 6.7% in the OCC, reflecting an effort to balance sensitivity with clinical specificity in GDM diagnosis [15].

The study is a secondary analysis of the randomised controlled trial (RCT), the LiP study [23], and the 3-year follow-up LiPO (Lifestyle in Pregnancy and Offspring) study [24].

The LiP study was conducted between October 2007 and October 2010 and included 360 pregnant women with pre-pregnancy Body Mass Index (BMI) ≥ 30 kg/m² from two Danish University Hospitals, Odense and Aarhus. The women were randomised to lifestyle intervention with healthy diet and physical activity from gestational age (GA) 10–14 and until delivery or to a standard care control group. Maternal weight, blood pressure, and a step test were assessed at baseline, at GA 35, and at six months postpartum. Based on the step test, VO₂max and cardiorespiratory fitness scores were estimated. Women completed questionnaires on demographics and mental health. OGTT and fasting blood samples were obtained at 12, 28, and 35 weeks of gestation, as well as at 6 months postpartum. A total of 21 women were excluded from the analysis because they met the Danish diagnostic criteria for GDM at some point during pregnancy and thus received treatment.

Following delivery, umbilical cord blood was collected, and neonatal weight and length were recorded. Birth weight was converted to Z-scores to assess deviations from population norms. Mode of delivery, including planned and emergency cesarean sections, was documented. Macrosomia was defined as ≥ 4000 g, and LGA or SGA as above the 90th or below the 10th percentile, based on Danish reference data. LGA and SGA were calculated using Marsal’s formula, incorporating fetal sex, birth weight, and gestational age [23, 25].

This analysis focuses exclusively on fasting glucose levels. Although it is possible that some participants had fasting glucose values above 5.6 mmol/L, they were not excluded unless they met the Danish 2-hour criterion (≥ 9.0 mmol/L) for GDM.

Fasting values are not included in the screening in Denmark and were not included at the time when the data were collected, and we do not use the 1-hour value, which was also not measured in our material.

The primary study was approved by the local ethics committee of the Region of Southern Denmark in 2007 (S-20070058) and was registered at ClinicalTrials.gov as NCT00530439 before inclusion of participants [23].

The LiPO study, conducted when the children were 2.5 to 3 years old, examined the effects of maternal obesity and prenatal lifestyle interventions on offspring body composition. Anthropometric measures included weight, height, skinfold thickness, and abdominal circumferences. Weight in light indoor clothing was recorded to the nearest 0.1 kg using a digital scale (Seca 704, Hamburg, Germany). Height was measured to the nearest 0.1 cm with a portable stadiometer (Seca 214). Triceps and subscapular skinfold thicknesses were assessed to the nearest 0.1 mm using a Harpenden caliper (Chasmors Ltd, London, UK). Abdominal circumference (at the umbilical level) and hip circumference (at the widest point of the buttocks) were measured to the nearest mm with a non-stretchable tape. Blood pressure was obtained in the supine position using an electronic device (Welch Allyn 420, Skaneateles Falls, NY, USA). All measurements were performed in triplicate and averaged [26]. Blood pressure was measured, and fasting blood samples were collected after a 4-hour fast for glucose, insulin, HDL, and triglycerides analysis. Dual-energy X-ray absorptiometry (DXA) scans were conducted to assess fat free mass, fat mass, and body fat percentage. DXA scans were conducted using a Lunar Prodigy scanner (GE Healthcare, Madison, WI, USA) equipped with ENCORE software (version 12.3, Prodigy; Lunar Corp, Madison, WI, USA). Due to the young age of the children, scan quality varied, with some being inadequate. Consequently, scans were categorized into four levels: (i) perfect, (ii) good with minor irregularities, (iii) several irregularities, and (iv) unusable. Scans graded as (iii) or (iv) were deemed unsatisfactory and excluded from further analyses. In the LiPO study, no significant differences in BMI Z-scores were observed between the intervention and control groups. A slight trend toward higher BMI Z-scores in the intervention group disappeared after adjusting for confounders [24].

We revalidated the DXA scans by thoroughly reviewing all previous scans. The fat percentage, fat mass, and fat free mass were derived from the DXA report and calculated by the DXA software.

Information on breastfeeding duration was collected using a self-administered questionnaire completed in the LiPO study. Mothers were asked to report the total duration of breastfeeding, which was recorded in predefined categories: <1 week, 1–2 weeks, 3–4 weeks, 1–2 months, 3–4 months, and > 6 months. Responses indicating “do not recall” were treated as missing data. For analyses of 3-year outcomes, breastfeeding duration was collapsed into two categories, representing breastfeeding for < 6 months and breastfeeding for ≥ 6 months, and was included as an interaction term with maternal fasting glucose to assess effect modification.

Statistical methods

Statistical analyses were carried out using Stata BE version 18.5 (StataCorp, College Station, TX, USA). The initial analyses evaluated maternal fasting plasma glucose (mmol/L) as a continuous exposure. Associations between fasting glucose (per 1 mmol/L increase) and a comprehensive set of offspring outcomes were examined using linear regression models. The outcomes included birth weight (grams), birth weight z-score (SD units), umbilical cord c-peptide (pmol/L), and at 3 years of age total fat mass (grams), total fat-free mass (grams), fat percentage, BMI z-score (SD units), HDL cholesterol (mmol/L), LDL cholesterol (mmol/L), HbA1c (mmol/mol), insulin (pmol/L), fasting glucose (mmol/L), c-peptide (pmol/L), systolic blood pressure (mmHg), and diastolic blood pressure (mmHg). Model assumptions were assessed using residual diagnostics, and regression coefficients with 95% confidence intervals were reported.

Effect modification by breastfeeding duration was assessed for 3-year outcomes by adding an interaction between fasting glucose and breastfeeding duration to the linear regression models.

Furthermore descriptive analyses were conducted, presenting the baseline characteristics of both mothers and offspring along with the variables of interest. Medians and IQRs were calculated for continuous variables and compared between groups using the Mann–Whitney U test.

We evaluated glucose cut-off levels ranging from 4.6 to 5.8 mmol/L in increments of 0.1 units (Figs. 1 and 2).

Fig. 1.

Neonatal birth outcomes categorized by maternal fasting glucose values

Fig. 2.

Outcomes at 3 years categorized by maternal fasting glucose values

For each GDM cut-off point, we analysed differences in numeric outcomes using linear regression and reported the mean differences between children with glucose levels below and above each cut-off. To analyse dichotomous outcomes, we conducted logistic regression analyses to estimate and report the OR associated with each glucose threshold, adjusting for maternal pre-pregnancy BMI and age.

Results

A total of 301 women completed the LiP study and were eligible for follow-up among these, 157 (56%) participated in the follow-up study, with a mean follow-up time of 2.8 years after delivery. DXA scans were conducted in 75 children. Higher maternal fasting plasma glucose at gestational week 28 was associated with a significantly higher birth weight z-score (β = 0.48, 95% CI 0.20 to 0.75, p = 0.001), while the association with birth weight in grams was not statistically significant (β = 103 g, 95% CI − 43 to 249, p = 0.167) (Table 1).

Table 1.

Associations between maternal fasting plasma glucose at GA 28 and offspring outcomes at birth and at 3 years

Outcome	Maternal Fasting Glucose at GA28 and Offspring Outcomes
	Coefficient	95% confidence intervals	P-value
Birth weight, grams	103	(−43–249)	0.167
Birth weight Z-score	0.48	(0.20–0.75)	0.001
C-peptid umbilical Cord, pmol/L	96.68	(−70.21-263.57)	0.253
3 years
Fat mass total, grams	160	(−0.30–0.62)	0.495
Fat free mass total, grams	262	(−623–1149)	0.167
Fat percent	0.90	(−1.92-3.71)	0.527
BMI z-score	−0.17	(−0.55–0.21)	0.385
HDL, mmol/L	0.01	(−0.12–0.13)	0.933
LDL, mmol/L	−0.01	(−0.27–0.25)	0.940
HbA1c, mmol/mol	−0.34	(−1.69-1.02)	0.624
Insulin, pmol/L	−5.93	(−12.98-1.12)	0.098
Fasting glucose, mmol/L	−0.04	(−0.23-0.15)	0.677
C-peptid pmol/L	−43.78	(−118.27-30.71)	0.246
Systolic BP, mmHg	−0.06	(−3.85–1.72)	0.451
Diastolic BP, mmHg	−1.82	(−4.00–0.37)	0.104

Linear regression results are shown as coefficients per 1 mmol/L increase in fasting glucose with 95% confidence intervals and p-values. Abbreviations: Body mass index (BMI), hemoglobin A1c (HbA1c), high-density lipoprotein (HDL), low-density lipoprotein (LDL), and blood pressure (BP)

At 3 years of age, maternal fasting glucose was not significantly associated with BMI z-score or other cardiometabolic measures. Nonetheless, a non-significant trend toward greater adiposity was observed, with higher maternal glucose relating to increased total fat mass (β = 160 g, 95% CI − 0.30 to 0.62, p = 0.495) and higher fat-free mass (β = 262 g, 95% CI − 623 to 1149, p = 0.167).

Across all outcomes, none of the interactions between breastfeeding duration and fasting glucose reached statistical significance, indicating no evidence of effect modification by breastfeeding duration (Table 2). This applied to measures of body composition, anthropometry, cardiometabolic markers, and blood pressure. Likewise, the main effects of maternal fasting glucose and breastfeeding duration were non-significant for 3-year outcomes.

Table 2.

Associations between maternal fasting plasma glucose, breastfeeding duration, and offspring cardiometabolic outcomes at 3 years of age, including interaction analyses

	Main effect: FPG (95% CI)	p-value	Main effect: Breastfeeding (95% CI)	p-value	Interaction: FPG x Breastfeeding b (95% CI)	p-value
Fat mass total, grams	−0.13 (−1.08-0.83)	0.788	−1.62 (−10.04-6.80)	0.697	0.28 (−1.43-2.00)	0.738
Fat free mass total, grams	−132 (−1924-1660)	0.881	−2454 (−18287-13379)	0.753	425 (−2799-3651)	0.789
Fat percent	−0.71 (−6.27-4.85)	0.796	−8.33 (−57.55-40.67)	0.727	1.53 (−8.47-11.54)	0.756
BMI z-score	−0.01 (−0.64-0.61)	0.968	3.71 (−2.32-9.74)	0.224	−0.74 (−1.97-0.50)	0.238
HDL, mmol/L	−0.06 (−0.24-0.12)	0.483	−0.86 (−2.66-0.95)	0.342	0.18 (−0.18-0.54)	0.313
LDL, mmol/L	0.05 (−0.38-0.49)	0.809	−2.21 (−6.57-2.16)	0.312	0.42/−0.46-1.29)	0.340
HbA1c, mmol/mol	−1.63 (−3.67-0.41)	0.115	−5.44 (−27.46-16.59)	0.620	0.99 (0.3.35–5.32)	0.648
Insulin, pmol/L	−5.81 (−18.22-6.60)	0.350	22.54 (−95.48-140.57)	0.702	−4.34 (−28.08-19.41)	0.714
Fasting glucose, mmol/L	−0.21 (−0.51-0.08)	0.154	−0.65 (−3.85-2.54)	0.681	0.12 (−0.51-0.75)	0.709
C-peptid pmol/L	−181.39 (−736.88-374.10)	0.508	−1163.63 (−5390.80-3063.53)	0.576	204.44 (−640.63-1049.52)	0.623
Systolic BP, mmHg	−4.21 (−8.73-0.31)	0.067	−28.83 (−69.71-12.06)	0.163	5.92 (−2.92-14.25)	0.160
Diastolic BP, mmHg	−3.77 (−7.95-0.40)	0.076	9.50 (−28.25-47.25)	0.616	−2.07 (−9.77-5.62)	0.591

Results from linear regression models examining the possible effect modification by breastfeeding duration on associations between maternal fasting plasma glucose (FPG) and offspring outcomes at 3 years of age. Estimates are presented as regression coefficients (β) with 95% confidence intervals (CI). Main effects are shown for maternal fasting plasma glucose (per 1 mmol/L increase) and breastfeeding duration (<6 months vs. ≥6 months)

Effect modification by breastfeeding duration was assessed by including an interaction term between FPG and breastfeeding duration (FPG × breastfeeding). P-values correspond to the main effects and the interaction term. No statistically significant interactions were observed

Abbreviations: Body mass index BMI, hemoglobin A1c HbA1c, high-density lipoprotein HDL, low-density lipoprotein LDL, and blood pressure BP

Results of the comparisons of both maternal and newborn outcomes between women who meet and those who do not meet the various GDM criteria are presented in Tables 3 and 4. Newborns to women diagnosed with GDM based on the WHO2013 criteria had higher birth weight standard deviation scores, Z-scores, compared to those born to women who did not meet the criteria (0.474 vs. 0.135, p = 0.01). Furthermore, in the group with fasting glucose levels above 5.6 mmol/L there was a higher proportion of large-for-gestational-age (LGA) infants. Newborns of mothers with fasting glucose levels above 5.3 mmol/L had higher c-peptide levels than those born to mothers with levels below 5.3 mmol/L (592 pmol/L vs. 430 pmol/L, p = 0.01) (Table 4).

Table 3.

Maternal baseline characteristics across three different diagnostic GDM thresholds

Maternal baseline characteristics	Fasting glucose ≥ 5.1 mmol/L			Fasting glucose ≥ 5.3 mmol/L			Fasting glucose ≥ 5.6 mmol/L
	GDM	No GDM	P-value	GDM	No GDM	P-value	GDM	No GDM	P-value
Number	69	207		57	219		18	258
Pre-pregnancy BMI, kg/m2	34.57 (32.40; 37.23)	33.12 (31.50; 35.86)	0.01	34.72 (32.38; 37.28)	33.12 (31.50; 35.86)	0.01	35.50 (33.81; 37.64)	33.19 (31.59; 36.36)	0.02
Age, years	30 (27; 33)	29 (26; 31)	0.04	30 (28; 33)	29 (26; 31)	0.03	29 (26; 30)	29 (26; 32)	0.46
GWG, kg	8.05 (5.45; 11.67)	7.75 (5.02; 10.87)	0.27	7.30 (5.25; 11.67)	7.95 (5.12; 10.90)	0.66	6.40 (4.00; 11.60)	7.90 (5.20; 11.10)	0.55

Maternal baseline characteristics across three different diagnostic GDM thresholds (fasting glucose ≥ 5.1, 5.3, and 5.6 mmol/L) during an oral glucose tolerance test at gestational age 28. Normally distributed continuous variables are presented as mean (SD), and differences between groups were assessed using Student’s t-test. Non-normally distributed continuous variables are reported as median with interquartile ranges [25%; 75%] and were compared using the Mann–Whitney U test. Categorical variables are expressed as frequencies (percentages), with group differences evaluated using the chi-square test or Fisher’s exact test, as appropriate. Abbreviations: Body mass index (BMI),Gestational weight gain (GWG)

Table 4.

Birth outcomes across three different diagnostic GDM thresholds

Birth offspring outcomes	Fasting glucose ≥ 5.1 mmol/L			Fasting glucose ≥ 5.3 mmol/L			Fasting glucose ≥ 5.6 mmol/L
	GDM	No GDM	P-value	GDM	No GDM	P-value	GDM	No GDM	P-value
Birth weight, grams	3792 (3419; 4113)	3644 (3366; 4040)	0.18	3792 (3419; 4113)	3650 (3366; 4040)	0.25	3836 (3406; 4260)	3672 (3380; 4042)	0.52
Birth weight z-score	0.474 (0.014; 1.099)	0.135 (−0.504; 0.668)	0.02	0.433 (0.014; 1.123)	0.151 (−0.483; 0.692)	0.02	0.753 (0.153; 1.697)	0.188 (−0.479; 0.740)	0.01
LGA, n	14 (20.3)	24 (11.6)	0.07	12 (21.1)	26(11.6)	0.07	6 (33.3)	30 (12.7)	0.01
SGA, n	4 (5.8)	14 (6.8)	0.78	2 (3.5)	16 (7.3)	0.30	0 (0.0)	18 (7.0)	0.25
NICU, n	12 (17.4)	21 (10.1)	0.11	11 (19.3)	22 (10.0)	0.06	4 (22.2)	29 (11.2)	0.17
AC, cm	34 (33; 35)	34 (32; 35)	0.92	34 (33; 36)	34 (32; 35)	0.70	34 (32; 36)	34 (32; 35)	0.98
Cord blood c-peptide, pmol/L	538 (400; 699)	436 (316; 549)	0.08	592 (438; 724)	430 (314; 545)	0.01	580 (457; 769)	440 (321; 558)	0.07

Birth outcomes across three different diagnostic GDM thresholds (fasting glucose ≥ 5.1, 5.3, and 5.6 mmol/L) during an oral glucose tolerance test at gestational age 28

Normally distributed continuous variables are presented as mean (SD), and differences between groups were assessed using Student’s t-test. Non-normally distributed continuous variables are reported as median with interquartile ranges [25%; 75%] and were compared using the Mann–Whitney U test. Categorical variables are expressed as frequencies (percentages), with group differences evaluated using the chi-square test or Fisher’s exact test, as appropriate

Abbreviations: Large for gestational age LGA, small for gestational age SGA, admission to the neonatal intensive care unit NICU, and abdominal circumference AC

Fat mass at 3 years of age did not differ significantly between children born to mothers who met the WHO2013 diagnostic criteria for GDM compared to those who did not. We found a trend toward higher BMI Z-scores, increased fat and fat free mass, and elevated cholesterol levels with rising maternal fasting glucose levels, albeit not statistically significant (Fig. 2; Table 5).

Table 5.

3 year offspring outcomes across three different diagnostic GDM thresholds (fasting glucose ≥ 5.1, 5.3, and 5.6 mmol/L) during an oral glucose tolerance test at gestational age 28

3 years offspring outcomes	Fasting glucose ≥ 5.1 mmol/L			Fasting glucose ≥ 5.3 mmol/L			Fasting glucose ≥ 5.6 mmol/L
	GDM	No GDM	P-value	GDM	No GDM	P-value	GDM	No GDM	P-value
Fat mass total, grams	2.017 (1.704; 2.349)	2.132 (1.505; 2.375)	1.00	2.075 (1.572; 2.308)	2.119 (1.533; 2.378)	0.93	2.267 (2.253; 3.856)	2.044 (1.526; 2.370)	0.11
Fat free mass total, grams	10.979 (10.196; 11.770)	11.256 (10.061; 12.233)	0.82	10.924 (10.052; 11.683)	11.272 (10.198; 12.188)	0.59	11.585 (10.979; 15.196)	11.233 (10.052; 12.146)	0.27
Fat percentage, %	19.15 (15.82; 20.28)	17.75 (15.53; 21.12)	0.87	19.18 (15.54; 20.32)	17.75 (15.69; 20.67)	0.96	20.17 (19.21; 24.94)	17.75 (15.58; 20.56)	0.16
BMI z-score	0.022 (−0.748; 0.501)	0.147 (−0.517; 0.824)	0.30	0.040 (−0.748; 0.501)	0.133 (−0.517; 0.824)	0.31	0.524 (−1.377; 0.725)	0.099 (−0.533; 0.756)	0.68
Glucose, mmol/L	6 (5; 6)	6 (5; 6)	0.50	6 (5; 6)	6 (5; 6)	0.60	5 (5; 6)	5 (5; 6)	0.51
HbA1C, mmol/mol	32 (31; 34)	32 (30; 34)	0.50	32(30; 33)	32 (30; 34)	0.60	31 (29; 35)	32 (31; 34)	0.51
Insulin, pmol/L	14 (9; 22)	12 (7; 23)	0.43	13 (8; 21)	12 (7; 23)	0.79	8 (5; 14)	13 (8; 23)	0.15
HOMA	3 (2; 5)	3 (2;6)	0.26	3 (2; 5)	3 (2; 5)	0.62	2 (1; 3)	3 (2; 5)	0.18
C-peptide, pmol/L	226 (156; 376)	200 (162; 289)	0.40	220 (157; 375)	200 (160; 289)	0.48	194 (129; 222)	209 (162; 301)	0.28
HDL, mmol/L	1.30 (1.10; 1.40)	1.25 (1.10; 1.40)	0.56	1.30 (1.10; 1.42)	1.30 (1.10; 1.40)	0.95	1.40 (1.25; 1.50)	1.30 (1.10; 1.40)	0.23
LDL, mmol/L	2.50 (2.00; 2.80)	2.30 (1.90; 2.68)	0.40	2.40 (1.98; 2.82)	2.30 (1.90; 2.62)	0.58	2.60 (1.90; 2.85)	2.40 (1.90; 2.60)	0.58
TC, mmol/L	4.00 (3.52; 4.67)	3.95 (3.50; 4.30)	0.72	3.95 (3.48; 4.80)	4.00 (3.50; 4.30)	0.80	4.30 (3.40; 4.90)	4.00 (3.50; 4.30)	0.56
Systolic BP, mmHg	97 (93; 103)	98 (94; 102)	0.45	97 (93; 105)	98 (94; 102)	0.50	100 (94; 106)	98 (94; 102)	0.48
Diastolic BP, mmHg	62 (60; 65)	64 (61; 67)	0.14	62 (59; 65)	64 (61; 67)	0.11	62 (58; 64)	64 (60; 67)	0.26

3 year offspring outcomes across three different diagnostic GDM thresholds (fasting glucose ≥ 5.1, 5.3, and 5.6 mmol/L) during an oral glucose tolerance test at gestational age 28. Normally distributed continuous variables are presented as mean (SD), and differences between groups were assessed using Student’s t-test. Non-normally distributed continuous variables are reported as median with interquartile ranges [25%; 75%] and were compared using the Mann–Whitney U test. Categorical variables are expressed as frequencies (percentages), with group differences evaluated using the chi-square test or Fisher’s exact test, as appropriate. Abbreviations: Body mass index (BMI), hemoglobin A1c (HbA1c), homeostatic model assessment for insulin resistance (HOMA), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), total cholesterol (TC), triglycerides (TG), and blood pressure (BP)

Discussion

In this study, maternal fasting plasma glucose was analysed as a continuous exposure, and we also examined how different diagnostic thresholds for GDM related to offspring metabolic outcomes at birth and at 3 years of age. Our findings indicate that neonatal outcomes were significantly affected by maternal GDM status. Specifically, offspring born to mothers who met the fasting WHO2013 criterion for GDM had significantly higher birth weight Z-scores. Furthermore, in the group with maternal fasting glucose levels above 5.6 mmol/L, there was a significantly higher proportion of LGA infants. Newborns of mothers with fasting glucose above 5.1 mmol/L also exhibited significantly higher C-peptide levels. Despite these findings at birth, no significant differences in fat mass or other anthropometric and metabolic parameters were observed at the three-year follow-up in children born to mothers who met the WHO2013 fasting criterion. These results align with numerous other studies reporting no early-life differences in anthropometric outcomes associated with GDM.

Our findings align with those of Mogensen et al. [27], who performed a secondary analysis of follow-up data from the APPROACH study to examine the impact of maternal glucose levels on offspring metabolic health. In their analysis of 208 women with overweight or obesity but without GDM, comparisons of maternal fasting and 2-h glucose levels during a 2-h OGTT in pregnancy showed no differences in offspring metabolic markers at 3 and 5 years of age.

These shared findings suggest that maternal hyperglycaemia may exert its strongest influence on neonatal outcomes, whereas associations during early childhood appear less consistent. Nevertheless, the absence of significant associations at three years of age in our study does not preclude the possibility that such associations may emerge later in development. The HAPO Family Study [28] likewise found no convincing effect of maternal glycaemia on obesity or adiposity at age two years.

These findings align with earlier evidence showing that metabolic profiles of offspring exposed to maternal diabetes are often indistinguishable from those of unexposed offspring at this young age. However, longer-term follow-up from this cohort and others, including findings from Associations between maternal and offspring glucose metabolism: a 9-year follow-up of a randomised controlled trial [29] has revealed emerging associations, indicating that early null findings do not rule out later effects.

Such studies reinforce the concept that intrauterine exposures may exert delayed or cumulative effects on metabolic function, detectable only as offspring progress into later childhood.

Longer-term follow-up may therefore reveal more subtle or cumulative effects of maternal glycaemia on obesity-related or dysmetabolic traits.

Insights from the Hyperglycaemia and Adverse Pregnancy Outcome Follow-up Study (HAPO FUS) illustrate how such associations may become evident at later developmental stages. In HAPO FUS, higher maternal glucose levels across the full continuum of the OGTT, even within non-diabetic ranges, were linearly associated with adverse metabolic outcomes in offspring aged 10–14 years [30]. Higher maternal fasting glucose predicted elevated offspring fasting glucose, higher HbA1c, and an increased risk of impaired fasting glucose. Higher maternal 1-hour and 2-hour glucose levels were associated with elevated glucose concentrations at all time points during the offspring OGTT and a greater risk of impaired glucose tolerance. Offspring exposed to higher maternal glycaemia also demonstrated lower insulin sensitivity and impaired β-cell compensation, reflected in a reduced disposition index. These relationships were independent of maternal BMI, child BMI, and family history of diabetes, supporting the hypothesis that intrauterine glycaemic exposure may have programming effects on later metabolic function.

In contrast, our study did not identify significant associations between maternal glucose levels and metabolic outcomes at age three years.

In addition, large-scale cohort studies have suggested that breastfeeding may attenuate adverse cardiometabolic profiles among children exposed to higher gestational fasting glycaemia. For example, Yi Ying Ong et al. [31] reported significant effect modification by breastfeeding in a cohort of 827 mother–child dyads, with protective associations evident at 6 years of age. In contrast, we did not observe evidence of effect modification by breastfeeding in interaction analyses at 3 years of age. This discrepancy may reflect limited statistical power to detect interactions in our relatively small sample, particularly for DXA-derived outcomes, as well as differences in outcome age, developmental stage, definition of breastfeeding exposure, and underlying population characteristics. Most notably, the developmental stage at which outcomes were assessed differs substantially between studies. The HAPO FUS offspring were evaluated in late childhood and early adolescence, a period characterised by pronounced physiological insulin resistance driven by pubertal hormonal changes. Puberty represents a critical metabolic window during which pre-existing vulnerabilities, including those originating from intrauterine glycaemic exposure, may be amplified. Evidence suggests that pubertal insulin resistance may fail to fully resolve in youth with underlying adiposity or adverse prenatal exposures, thereby unmasking metabolic disturbances not evident earlier in life [32].

This mechanism is further supported by findings from the EPICOM study, in which adolescent offspring of mothers with type 1 diabetes exhibited reduced insulin sensitivity and disposition index at 17–18 years of age, particularly among females, despite more modest differences earlier in childhood [33].

Taken together, findings from APPROACH follow-up, HAPO FUS, EPICOM, Nyen et al., and our own study indicate that the metabolic consequences of maternal hyperglycaemia may be subtle or inconsistent in early childhood but become increasingly apparent later in development, particularly during or after puberty.

The absence of significant associations at three years of age may reflect the influence of postnatal environmental factors that shape growth and metabolic development. Early dietary practices, and general family lifestyle may mitigate or enhance early metabolic risk. Multiple postnatal exposures may have opposing effects, and these influences may reduce or obscure associations at this age. Early metabolic differences are often inconsistent in infancy and early childhood, and they may become more pronounced later in development, particularly during puberty. A wide range of environmental and behavioural factors, including diet, physical activity, and family lifestyle, may contribute to early growth trajectories. Continued longitudinal follow-up of this cohort is needed to determine whether maternal glycaemia contributes to later differences in insulin sensitivity, beta cell function, adiposity, or cardiometabolic health.

Continued longitudinal follow-up into later childhood and adolescence will be essential to determine whether maternal glycaemia contributes to divergence in insulin sensitivity, β-cell function, or glucose tolerance over time. In this context, it is also relevant to consider findings from large lifestyle intervention trials conducted during pregnancy, including LIMIT [34], UPBEAT [35] and RADIEL [36]. Although these trials were not designed primarily to investigate gestational glycaemia or offspring metabolic patterns, their results provide important complementary evidence. Across these studies, lifestyle interventions generally did not lead to substantial improvements in neonatal or early childhood anthropometric or metabolic outcomes, mirroring the pattern observed in our cohort and in other early-life follow-ups such as Mogensen et al. and the HAPO Family Study [28].

These convergent findings support the notion that early childhood may represent a developmental period during which metabolic differences arising from intrauterine glycaemic exposure are subtle or not yet detectable.

Several of these trials, most notably RADIEL [36], did, however, demonstrate reductions in the incidence of GDM, indicating that maternal glycaemia is at least partially modifiable. This provides indirect mechanistic support for the hypothesis that intrauterine glycaemic exposure may influence later metabolic outcomes. If GDM and maternal hyperglycaemia can be reduced through lifestyle modification, the downstream metabolic consequences observed in later childhood and adolescence may also be preventable.

One explanation for the limited effects on offspring outcomes in these trials may therefore be that improved maternal glycaemic control in the intervention groups reduced fetal exposure, which in turn attenuated differences detectable in infancy or early childhood. In this way, these studies reinforce the concept that maternal glycaemia is a modifiable intrauterine pathway, even if immediate offspring phenotype does not markedly differ in early life.

Furthermore, the comprehensive and rigorously collected datasets from these multicentre trials represent a unique opportunity for secondary analyses. These datasets allow researchers to examine the same exposures investigated in the present study, namely maternal glucose homeostasis, within well-characterised populations, thereby strengthening the evidence that fetal metabolic programming may only become apparent later in development.

A key strength of this study is the robust and detailed data collection within an RCT, along with the unique opportunity to examine untreated GDM based on fasting values.

The unique dataset also allows for exploration of untreated GDM based on fasting values, reflecting the specificity of Danish diagnostic criteria. However, the study has limitations, notably the low follow-up rate at the three-year assessment, with approximately 50% participation. Additionally, some children were unable to complete DXA scans, resulting in missing data and potentially reducing the generalisability. The relatively small sample sizes in certain subgroups may have limited statistical power, increasing the risk of a type II error. Therefore, the lack of significant findings in some outcomes should be interpreted with caution.

Women who meet the GDM criteria according to WHO2013 but not the Danish diagnostic criteria may represent an underdiagnosed cohort that could benefit from early intervention [13]. Further investigation into this subgroup is necessary to determine whether targeted management could reduce adverse outcomes for both mothers and their offspring, not only during pregnancy but also in the long term. Indeed, the long-term consequences of GDM include a significantly increased maternal risk for developing type 2 diabetes, with affected women facing nearly a 10-fold higher risk compared to those with normoglycemic pregnancies [37]. Women with prior GDM are at increased risk of complications in subsequent pregnancies and postpartum weight retention, raising the likelihood of obesity and related conditions such as hypertension, hyperlipidaemia, and metabolic syndrome [38–40]. While this association is well documented in mothers, the impact on their offspring remains less thoroughly investigated. Long-term follow-up studies, such as the ongoing LiPO-Teen project (clinicaltrials.gov, NCT05774652, 03/21/2023), provide a unique opportunity to address these important questions. The LiPO-Teen study aims to evaluate whether lifestyle interventions during pregnancy improve health outcomes in both mothers and their children, with a focus on the long-term effects 15 years after the intervention. The primary aim of the study is to assess the impact of prenatal lifestyle interventions on offspring body composition at 15 years of age measured through DXA scanning. Furthermore, it provides a framework to assess whether maternal hyperglycemia contributes to long-term metabolic risks in offspring. Hence, there is a need for further and larger studies to confirm and expand on these findings to inform clinical practice and public health strategies. Such work will be essential for refining GDM diagnostic criteria, improving clinical care, and informing interventions to address the intergenerational cycle of obesity and metabolic disorders.

Conclusion

This study highlights the significant associations between maternal fasting glucose levels and neonatal outcomes, such as birth weight and c-peptide levels, while showing no differences in offspring fat mass at 3 years of age. These findings are consistent with previous research suggesting that the effects of maternal hyperglycemia may be most pronounced at birth and less evident in early childhood. However, the potential for long-term effects of maternal glucose levels on offspring metabolic health remains unclear.

Abbreviations

AC

Abdominal circumference

BMI

Body Mass Index

BP

Blood pressure

GA

Gestational age

GDM

Gestational diabetes mellitus

GWG

Gestational weight gain

HbA1c

Hemoglobin A1c

HDL-c

High-density lipoprotein cholesterol

HOMA

Homeostatic model assessment for insulin resistance

HAPO

Hyperglycemia and Adverse Pregnancy Outcome

HAPO FUS

Hyperglycemia and Adverse Pregnancy Outcome Follow-up Study

LGA

Large-for-gestational-age

LiP

Lifestyle in Pregnancy (Intervention in Pregnancy)

LiPO

Lifestyle in Pregnancy and Offspring

LDL-c

Low-density lipoprotein cholesterol

NICU

Admission to the neonatal intensive care unit

OCC

Odense Child Cohort

OGTT

Oral glucose tolerance test

RCT

Randomised controlled trial

SGA

Small for gestational age

TC

Total cholesterol

TG

Triglycerides

Authors’ contributions

C.A.V., M.H.T., J.S.J., H.T.C., and D.M.J. designed and conducted the original LiP and LiPO studies. For this secondary analysis, B.M.L., L.L.S., M.O., and D.M.J. developed the concept and study design. B.M.L., L.L.S., D.M.J., S.M., and C.A.V. analyzed and interpreted the data, and B.M.L. drafted the manuscript and prepared tables and figures. S.M., B.M.L., and L.L.S. performed the statistical analyses. During manuscript preparation, ChatGPT was used for minor language editing but not for conceptual development, study design, or data analysis. All authors (B.M.L., L.L.S., M.H.T., J.A., S.M., P.G.O., H.T.C., H.D.M., P.M.C., J.S.J., M.O., D.M.J., and C.A.V.) critically revised the manuscript and approved the final version for publication.

Funding

Open access funding provided by University of Southern Denmark. This study received no specific funding.

Data availability

The data that support the findings of this study are available from C.A.V., but restrictions apply to their availability. The data were used under license for the current study and include unpublished results that are still being analyzed. Therefore, they are not publicly available. However, data may be obtained from the corresponding author on reasonable request and with permission from C.A.V.

Declarations

Ethics approval and consent to participate

Both the LiP and LiPO studies received approval from the Regional Ethics Committee of Southern Denmark (S-20070058) and were registered with the Danish Data Protection Agency. The LiP study was registered at www.clinicaltrials.gov (NCT00530439) with a registration date of September 13, 2007. The LiPO follow-up study, initiated in 2010 while the LiP study was still ongoing, was registered at www.clinicaltrials.gov (NCT01918319) to investigate the offspring of LiP participants. All participants provided informed, written consent prior to their inclusion in the studies, in accordance with the Declaration of Helsinki and applicable national guidelines.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Authors’ information

Not applicable.