Background
Maternal obesity increases the risk for pregnancy complications and adverse neonatal outcome and has been associated with long-lasting adverse effects in the offspring, including increased body fat mass, insulin resistance, and increased risk for premature cardiovascular disease. Lifestyle interventions in pregnancy have produced no or modest effects in the reduction of adverse pregnancy outcomes in obese mothers. The Metformin in Obese Pregnant Women trial was associated with reduced adverse pregnancy outcomes and had no effect on birthweight. However, the long-term implications of metformin on the health of offspring remain unknown.
Objective
The purpose of this study was to assess whether prenatal exposure to metformin can improve the cardiovascular profile and body composition in the offspring of obese mothers.
Study Design
In 151 children from the Metformin in Obese Pregnant Women trial, body composition, peripheral blood pressure, and arterial pulse wave velocity were measured. Central hemodynamics (central blood pressure and augmentation index) were estimated with the use of an oscillometric device. Left ventricular cardiac function and structure were assessed by echocardiography.
Results
Children were 3.9±1.0 years old, and 77 of them had been exposed to metformin prenatally. There was no significant difference in peripheral blood pressure, arterial stiffness, and body composition apart from gluteal and tricep circumferences, which were lower in the metformin group ( P <.05). The metformin group, compared with the placebo group, had lower central hemodynamics (mean adjusted decrease, –0.707 mm Hg for aortic systolic blood pressure, –1.65 mm Hg for aortic pulse pressure, and –2.68% for augmentation index; P <.05 for all) and lower left ventricular diastolic function (adjusted difference in left atrial area, –0.525 cm 2 , in isovolumic relaxation time, –0.324 msec, and in pulmonary venous systolic wave, 2.97 cm/s; P <.05 for all). There were no significant differences in metabolic profile between the groups.
Conclusion
Children of obese mothers who were exposed prenatally to metformin, compared with those who were exposed to placebo, had lower central hemodynamic and cardiac diastolic indices. These results suggest that the administration of metformin in obese pregnant women potentially may have a beneficial cardiovascular effect for their offspring.
Maternal obesity increases the risk for pregnancy complications and adverse neonatal outcomes and may have long-lasting adverse effects in the offspring, such as increased body fat mass and systolic blood pressure (SBP) in childhood, increased insulin resistance and dyslipidemia both in childhood and young adulthood, and increased risk for premature all-cause death and hospital admissions for cardiovascular events.
Why was this study conducted?
Maternal obesity is associated with an adverse cardiometabolic outcome in the offspring. The purpose of this study was to assess whether in utero exposure to metformin can impact on cardiometabolic profile and body fat distribution in the offspring of obese mothers who participated in the Metformin in Obese Pregnant Women randomized controlled trial.
Key findings
Children of obese mothers who were exposed to metformin in utero had improved central hemodynamics and left ventricular diastolic functional indices. No harmful effect on body composition was noted.
What does this add to what is known?
The results of the study suggest that metformin has a beneficial effect on the cardiovascular system of the offspring of obese mothers. The clinical implications of this finding require further exploration.
Randomized controlled trials on overweight and obese women during pregnancy have investigated the effect of interventions in the reduction of adverse pregnancy outcomes; however, very few trials have reported the influences of these interventions on the health of the offspring. Following dietary and lifestyle interventions in obese mothers, pregnancy outcomes have been largely unaffected, and changes in body fat distribution of the offspring have been none or only modest. Pharmacologic interventions might produce a greater response; to date, few trials have examined the effect of metformin in obese nondiabetic women in the reduction of adverse pregnancy outcomes. Although the primary outcome (birthweight) was similar between groups in both studies, experimental and clinical data suggest that in utero exposure to metformin use can have long-term effects in the offspring by modifying processes that regulate fat accumulation and cardiovascular health.
To investigate this hypothesis, we followed children from the Metformin in Obese Pregnant Women trial to assess whether in utero exposure to metformin can improve the cardiometabolic profile and body fat distribution in the offspring of obese mothers.
Methods
Study population and study design
Our study population consisted of the offspring from the Metformin in Obese Pregnant Women trial; in this trial, obese (body mass index, >35 kg/m 2 ) nondiabetic pregnant women were assigned randomly to receive metformin or placebo from 12–18 weeks gestation until delivery in 3 National Health Service maternity hospitals in the United Kingdom. In this study, we aimed to invite the mothers of the 393 cases with live births to bring their child to the Harris Birthright Research Centre for detailed cardiometabolic phenotyping. The examinations were conducted by 1 trained clinical research fellow (O.P.) who was blinded to all maternal information, including arm of randomization.
Ethical approval for the study was obtained from the London-Surrey Borders Research Ethics Committee (REC no 08/H0806/80). Signed informed consent was obtained from the parents and assent from the child when possible.
Adiposity measures
The following measurements were recorded while children were standing with arms hanging down to the side: (1) weight and height, (2) arm relaxed, arm flexed and tense, waist, gluteal, mid-thigh and calf circumferences with the use of a flexible tape with 0.5 cm width and 0.5 mm precision, and (3) skinfold thickness at the biceps, triceps, subscapular, supraspinal, and medial calf with the use of a calibrated Harpenden caliper (Baty International, West Sussex, UK) according to the International Society for Advancement of Kinanthropometry. All anthropometric measurements were performed in duplicate, and the mean of the measurements is provided.
Body fat distribution was determined as previously reported with the BIA-ACC device (BioTekna, Inc, Venice, Italy), , with the children dressed in light clothing without wearing any shoes. Information about weight gain since birth was obtained by measurements recorded by health visitors in the Child’s Health Record (Red book).
Hemodynamic measurements and vascular measurements
Peripheral SBP and diastolic blood pressure were measured as the average of the last 2 seated readings with an automated oscillometric device (Welch Allyn spot vital signs; Welch Allyn, Skaneateles Falls, NY) in the right arm with the use of the appropriate sized-cuff after a 5-minute rest. Carotid-to-femoral pulse wave velocity was measured with the Vicorder device (software version 4.0; Skidmore Medical Limited, Bristol, United Kingdom). The method has been previously described and has excellent intra- and interobserver repeatability and ease of use in childhood. The device also determines brachial oscillometric blood pressure with the use of a cuff that is placed around the upper arm. Central blood pressure parameters (aortic SBP and pulse pressure and augmentation index) are then derived from brachial blood pressure waveforms by the application of a previously described brachial to aortic transfer function.
Measures of cardiovascular function and structure
Conventional and tissue Doppler echocardiography was performed with the use of a Philips CX50 system (Philips Healthcare, Netherlands) according to American Society of Echocardiography guidelines. Measures which were assessed included left ventricular mass (LVM) and relative wall thickness, measures of systolic and diastolic function:peak systolic mitral annular tissue velocity, and midwall fractional shortening and peak mitral annular velocities in early diastole (e’), a measure of diastolic relaxation. The ratio of early diastolic transmitral flow velocity E/e’ was calculated. Left atrial area was measured in the apical 4 chamber view at the ventricular end systole. LVM measurements were normalized to height 2.7 , as indexed LVM. LVM Z-scores were calculated for all children. Global strain analysis included the average of all 16 segments, and the peak systolic strain values were reported by the use of 2-dimensional speckle-tracking software (QLAB, version 9.0, Philips Healthcare, Andover, MA). Right ventricle systolic function was also assessed by tricuspid annular plane systolic excursion. All measurements were performed by the same clinical Fellow who was trained in pediatric echocardiography.
Biomarker analysis
A 5-mL nonfasting venous blood sample was taken according to standard procedures for children whose parents have agreed to venipuncture. A numbing cream was applied 30 minutes before venipuncture to minimize discomfort. Serum lipids (total cholesterol, triglycerides, and high-density lipoprotein cholesterol and low-density cholesterol) were measured by modification of the standard Lipid Research Clinics Protocol with the use of enzymatic reagents for lipid determination. All assay coefficients of variation were <5%. High sensitivity C-reactive protein and leptin and adiponectin were measured with enzyme-linked immunosorbent assay methods. All samples were separated and frozen at –80°C within 1 hour of collection.
Statistical analysis
Continuous variables are expressed as mean±standard deviation (SD) or median and interquartile range (IQR) if not following the normal distribution. Numeric variables are presented as number (percentage). Normality of distribution was evaluated graphically by histograms and Q-Q plots. Inverse rank normalization was used to allow for unbiased estimates of effect sizes in regression analysis of dependent cardiometabolic parameters that deviated from Gaussian distribution.
Comparison of anthropometric, hemodynamic, and cardiometabolic parameters in offspring of mothers who were exposed to metformin or placebo in pregnancy was performed with the use of independent samples t-test or the nonparametric Mann-Whitney test and chi-squared test. Subsequently, we used multivariable linear regression analysis to identify independent determinants of the cardiometabolic profile of the offspring. Adiposity measures and cardiovascular measures with a signal of difference ( P <.1) in unadjusted comparisons among groups were used as outcome variables. Independent variables in regression models were prespecified on the basis of biologic plausibility and observed differences between the total randomized population of the Metformin in Obese Pregnant Women trial and the current sample, and no selection procedure was followed. The use of metformin in pregnancy was inserted in all models as a factor variable, and its effect size was adjusted for available exposure variables. In detail, multivariable regression models for hemodynamic and cardiac outcomes included offspring age, mother’s age at conception, sex, weight, height, race, blood pressure indices, and heart rate. For adiposity outcomes, we used the same set of confounders. Possible collinearity among covariates in regression analysis was explored through assessment of the variance inflation factor. We compared nested multivariable models with and without appropriate terms (likelihood ratio tests) to assess potential effect modification of metformin by sex. In certain analyses of dichotomous outcomes (ie, highest vs lower tertiles of transmitral flow ratio), adjusted logistic regression analysis was used.
Multilevel linear mixed model analysis was used to examine the impact of metformin exposure during pregnancy on longitudinal measurements of weight gain of the offspring across a range of approximately 7 years. The linear mixed model analysis included 2 random effects (random slope and random intercept) with unstructured variance-covariance and was adjusted for time intervals between sequential measurements, exposure to placebo or metformin, gender, race, changes in height, and mother’s conception age as fixed effects.
Statistical analysis was performed with the Stata software package(version 13.1; StataCorp, College Station, TX). All tests were 2-sided, and statistical significance was P <.05.
Our sample size of 151 subjects, allocated in 2 unequal groups of 77 and 74 participants, provided power of 80% to detect a clinically significant difference of 0.5×IQR in weight and/or 0.5×SD in SBP between children who were exposed to metformin and placebo, respectively. The dispersion parameters of SD and IQR for power calculations were retrieved from previous published data. Power analysis for the nonparametric Mann-Whitney test was based on 2000 simulations with resampling.
Results
Study population
In the Metformin in Obese Pregnant Women trial cohort, there were 393 live births; 86 live births (11.8%) were not contactable. Of the 307 women (78.2%) who were invited to participate in this study, 156 women (50.8%) refused. In total, 151 children (38.5%) were assessed, including 77 from the metformin and 74 from the placebo groups. Compared with the total Metformin in Obese Pregnant Women trial cohort, mothers in the current study were older (32.8 ±5.2 years vs 31.4 ±5.8; P<.01); however, no other differences in the risk factor profile and in the incidence of pregnancy complications were noted ( Supplemental Table ). The characteristics of the current study population are described in Table 1 . As previously shown, obese mothers who were treated with metformin during pregnancy had reduced gestational weight gain and incidence of preeclampsia, compared with women who received placebo.
| Characteristic | Placebo (N=74) | Metformin (N=77) | P value a |
|---|---|---|---|
| Gender of offspring (male), % | 50 | 49.4 | .94 |
| Age at follow up, y b | 37±5.2 | 37.5±5.8 | .60 |
| Race, n (%) | .86 | ||
| White | 44 (59.5) | 49 (63.6) | |
| Afro-Caribbean | 25 (33.8) | 23 (29.9) | |
| Asian | 5 (6.8) | 5 (5.5) | |
| Weight at 12 weeks gestation, kg c | 106 (97-121) | 104 (93 -114) | .17 |
| Body mass index at 12 weeks gestation (kg/m 2 ) b | 40±4.8 | 39.6±5.1 | .67 |
| Gestational weight gain, kg c | 7.1 (4.2-9.6) | 3.7 (1.2-7) | <.001 |
| Smoking, n (%) | 4 (5.4) | 5 (6.5) | .78 |
| Preeclampsia, n (%) | 6 (8.1) | 1 (1.3) | <.05 |
| Gestational diabetes mellitus, n (%) | 12 (16.2) | 13 (16.9) | .91 |
a Derived from independent samples t-test or Mann-Whitney test and chi-squared test
b Data are presented as mean±standard deviation
Adiposity phenotype
Children in the metformin group, compared with the placebo group, had no significant difference in weight, height, body mass index, skinfold, and body fat distribution measurements but had lower gluteal circumference (56.5 vs 58.3 cm; P <.05) and tricep circumference (30.2 vs 31.4 cm; P <.05; Table 2 ). After adjustment for age, sex, race, weight, and height, metformin exposure was associated independently with decreased gluteal circumference (mean difference after inverse rank normalization, –0.183; 95% confidence interval, –0.344 to –0.022; P <.03) and tricep circumference (mean difference, –0.189; 95% confidence interval, –0.376 to –0.001; P <.05). The rate of weight gain from birth to early childhood was also comparable between the 2 groups ( P =.579 for interaction of time×group classification; ie, metformin or placebo; Supplemental Figure ).
| Variable | Placebo | Metformin | P value a |
|---|---|---|---|
| Metabolic profile b | |||
| Cholesterol, mmol/L c | 3.7±0.64 | 3.8±0.66 | .32 |
| Low-density lipoprotein, mmol/L c | 1.5±0.33 | 1.5±0.33 | .43 |
| High-density lipoprotein, mmol/L c | 2.3±0.54 | 2.4±0.60 | .41 |
| Triglycerides, mmol/L d | 0.91 (0.69–1.38) | 0.94 (0.74–1.73) | .74 |
| Non high density lipoprotein, mmol/L d | 2.47 (2.13–2.93) | 2.70 (2.25–2.96) | .31 |
| C-reactive protein, mg/L d | 0.39 (0.22–1.25) | 0.62 (0.27–2.22) | .50 |
| Adiponectin, mg/L d | 13.30 (9.81–15.48) | 13.10 (11.60–14.82) | .60 |
| Leptin, μg/L d | 2.50 (1.70 –5.78) | 2.07 (1.53–2.93) | .18 |
| Body composition e | |||
| Birthweight, kg d | 3.5 (3.1–3.7) | 3.32 (3.0–3.68) | .28 |
| Weight, kg d | 18.7 (15.6–21.2) | 17.3 (15.7–20.1) | .15 |
| Height, m d | 1.04 (0.99–1.09) | 1.02 (0.97–1.07) | .25 |
| Body mass index, kg/m 2 c | 17.4±2.1 | 17.0±2.0 | .27 |
| Waist circumference, cm d | 52.3 (50.0–55.2) | 51.3 (49.4–53.8) | .12 |
| Gluteal circumference, cm d | 58.3 (54.5–62.0) | 56.5 (53.5–59.5) | .02 |
| Triceps circumference, cm d | 31.4 (29.0–34.7) | 30.2 (27.8–32.7) | .04 |
| Calf circumference, cm d | 22.5 (21.1–24.2) | 21.9 (20.4–23.3) | .18 |
| Triceps skinfold, mm d | 11.3 (9.2–13) | 10.8 (9.0–12.6) | .44 |
| Biceps skinfold, mm d | 5.6 (4.9–6.8) | 5.2 (4.4–6.2) | .10 |
| Subscapularis skinfold, mm d | 6.7 (5.8–8.7) | 6.7 (5.7–8.7) | .77 |
| Supraspinal skinfold, mm d | 5.8 (4.8–7.3) | 5.8 (4.4–7.0) | .24 |
| Medial calf skinfold, mm d | 11.1 (9.1–13.7) | 11.1 (9.5–12.7) | .80 |
| Free fat mass, kg d | 16.2 (14.1–18.2) | 15.6 (14.0–17.2) | .17 |
| Fat mass, kg d | 2 (1.1–3.3) | 2.2 (1.2–2.8) | .92 |
| Maximum oxygen uptake d | 48.1 (44.2–53.4) | 48.6 (45.4–51.6) | .75 |
| Total body water, L d | 14.1 (12.2–16.1) | 14.0 (12.0–15.7) | .27 |
| Extracellular water, L d | 7.6 (7.2– 7.9) | 7.4 (7.1–7.8) | .11 |
| Intracellular water, L d | 6.4 (4.8–8.2) | 7.0 (4.7–8.1) | .47 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
