Академический Документы
Профессиональный Документы
Культура Документы
12964
www.bjog.org
Objective To compare the growth and development of children Results Children exposed to metformin were significantly heavier
born to mothers with gestational diabetes mellitus (GDM) (10.47 versus 9.85 kg, 95% CI 0.04–1.20) at the age of 12 months
requiring pharmacological treatment, and randomised to and taller and heavier (83.9 vs 82.2 cm, 95% CI 0.23–3.03, 12.05
treatment with metformin or insulin. vs 11.32 kg, 95% CI 0.04–1.43) at the age of 18 months. The
mean ponderal index (PI) did not differ significantly. The motor,
Design Follow-up of a randomised controlled trial (RCT)
social, or linguistic development evaluated at the age of
comparing metformin and insulin treatment of GDM.
18 months did not differ between the groups.
Setting Data were gathered during routine visits to child welfare
Conclusions Children prenatally exposed to metformin were
clinics at the ages of 6, 12, and 18 months, including weight and
heavier at the 12 months measurements and taller and heavier at
height measurements, and assessment of motor, social, and
the 18 months measurements than those exposed to insulin, but
linguistic development.
their body composition defined by PI did not differ. Over the
Sample The children of mothers with GDM randomised to short term, metformin does not seem to be harmful with regards
metformin (n = 47) or insulin (n = 50) treatment during to early motor, linguistic, or social development.
pregnancy.
Keywords Gestational diabetes mellitus, metformin, offspring.
Methods Data were collected from the structured questionnaire
Linked article This article is commented on by Thom EA. To
filled in at the child welfare clinics.
view this mini commentary visit http://dx.doi.org/10.1111/
Main outcome measures The growth and development of the 1471-0528.13092.
children until the age of 18 months.
Please cite this paper as: Ijäs H, Vääräsmäki M, Saarela T, Keravuo R, Raudaskoski T. A follow-up of a randomised study of metformin and insulin in
gestational diabetes mellitus: growth and development of the children at the age of 18 months. BJOG 2014; DOI: 10.1111/1471-0528.12964.
harmful for the offspring over the short term, little is head circumference at the ages of 6 and 12 months from
known about its long-term effects. the clinic records. The check-up at 18 months of age
The Metformin in Gestational diabetes (MiG) study is included: an examination of vision, with an examination of
the largest randomised trial comparing metformin with corneal light reflex and red reflex using an ophthalmo-
insulin treatment in GDM.3 The MiG Offspring Follow-Up scope; Hirschberg and cover tests, in the assessment of stra-
(MiG TOFU) study at 2 years of age demonstrated no dif- bismus; and pinch grip, which tests hand–eye coordination.
ference in the overall growth between the children exposed The evaluation of motor functions included the capability
to either metformin or insulin in utero, although there was of standing and walking without support, normal pinch
a tendency to a more favourable pattern of fat distribution grip, and coordination of upper limbs. The social, emo-
in the children exposed to metformin.14 tional, and language development of the child was evalu-
We previously conducted an RCT to compare metformin ated by testing the response to spoken commands, and the
and insulin as a pharmacological treatment of GDM.4 In ability to make understandable spoken words and recipro-
our study these treatments were found to be equal in the cal contact. The response rate of questionnaires was 96%,
prevention of fetal macrosomy, and the incidence of as the questionnaire was returned by 45/47 mothers in the
short-term perinatal adverse events did not differ between metformin-treated group, and by 48/50 mothers in the
the study groups. The aim of the present study was to insulin-treated group (Figure 1).
compare the growth as well as the motor, social, and lin- The weight and height percentiles of the children at dif-
guistic development of the children prenatally exposed to ferent ages were calculated using the recently updated Finn-
metformin or insulin up to the age of 18 months. ish growth reference software.15 The number of children
growing above the 95–percentile or below the 5–percentile
lines was recorded. Weight-for-length (i.e. difference in
Methods
percentage to the median weight of the children of the
Our RCT study protocol and pregnancy outcome has been same length) was calculated by using the same software.
described in detail previously.4 Briefly, 100 pregnant According to Finnish guidelines for childhood obesity, a
women with GDM were randomised to receive either met- weight-for-length percentage of 10–20% defines overweight,
formin (n = 50) or insulin (n = 50) therapy when diet and a weight-for-length percentage of over 20% defines
turned out to be insufficient to achieve normoglycaemia. In obesity.16 To evaluate body composition, ponderal index
the metformin group, two women withdrew from the study (PI) was calculated as weight in kg divided by height in m3
shortly after randomisation, and one woman was excluded [1 000 000 9 (weight in kg)/(height in cm)³].
because of abnormal liver function tests. Out of the group The primary outcome was weight and height of the
of 47 in the metformin group, 15 (32%) needed supple- children at the ages of 6, 12, and 18 months. The second-
mental insulin to achieve normoglycaemia. One woman in ary outcomes included motor, social, and linguistic
the study population was of Asian origin, and the rest were development of the children until the age of 18 months, as
white. The study was carried out in the tertiary-level Oulu well as the body composition of the children defined by PI,
University Hospital, Oulu, and secondary-level Kainuu and the incidence of overweight and obesity defined by
Central Hospital, Kajaani. The study was approved by the weight-for-length.
Ethics Committee of Northern Ostrobothnia Hospital Dis-
trict and the Finnish National Agency of Medicines.
In Finland, free primary health care is offered to all Study period: 2007-2011
mothers and children at maternal and child welfare clinics,
Metformin n = 50 Insulin n = 50
with practically 100% attendance. Until the age of 6 years, Withdrawn
children are offered a schedule of check-ups, where stan- n=3
dardised weight and height measurements are performed
and the stage of development of the child is evaluated by a Metformin only Metformin +supplemental
n = 32 insulin n = 15
general practitioner and/or a nurse specially trained in
child health care.
The mothers of this study returned a questionnaire Not returned Not returned Not returned
the questionnaire the questionnaire the questionnaire
designed by a paediatrician (T. Saarela). The questionnaire n=1 n=1 n=2
was filled in by a specially trained nurse at the child welfare
clinic at the check-up at 18 months of age, including data Offspring follow-up at Offspring follow-up Offspring follow-up at
on height, weight, and head circumference, as well as the age of 18 months at the age of 18 months the age of 18 months
n = 31 n = 14 n = 48
motor, social, and linquistic development, examined by the
doctor. The nurse also filled in data on height, weight, and Figure 1. Flowchart.
Mothers
Age at randomisation (years) 32.1 5.5 31.9 6.2 0.872
Pre-pregnancy BMI (kg/m²) 31.0 6.2 30.6 5.4 0.744
Smoking 1 (2.2) 4 (8.3) 0.201 0.27 0.03–2.30
Gestational age at randomisation (weeks) 30.0 4.5 30.4 4.1 0.822*
GhbA1c at randomisation (%) 5.9 0.41 5.9 0.40 0.674
Gestational age at delivery (weeks) 39.3 1.2 38.9 4.1 0.355*
Infants
Male/female 21 (47)/24 (53) 22 (46)/26 (54)
Birthweight (g) 3694 426 3528 585 0.119
Height (cm) 50.8 1.8 49.9 2.2 0.047
PI (kg/m³) 28.4 2.4 28.6 3.0 0.710
Head circumference (cm) 35.2 1.4 35.2 1.7 0.888
LGA n (%)* 3 (6.7) 4 (8.3) 1.000 0.8 0.19–3.38
Data are means SDs (Student’s t–test/*Mann–Whitney U–test) or n (%) (Fisher’s exact test).
[Correction added on 3 December 2014 after first online publication: the statistical data in Table 1 were miscalculated; and these have been
corrected.]
6 months
Height (cm) 68.1 3.8 67.4 2.7 0.286 0.63–2.11*
≥95 percentile 1 (2.2) 0 (0) 0.484
≤5 percentile 0 (0) 8 (16.7) 0.006
Weight (kg) 8.281 0.99 7.925 0.99 0.071 0.03–0.79
≥95 percentile 4 (8.9) 3 (6.3) 0.709 1.42 0.34–6.00
≤5 percentile 1 (2.2) 4 (8.3) 0.363 0.27 0.03–2.30
Weight-for-length > 10% 9 (20.0) 8 (16.7) 0.790 1.20 0.51–2.84
Weight-for-length > 20% 3 (6.7) 2 (4.2) 0.671 1.60 0.28–9.14
PI (kg/m³) 26.8 8.0 25.8 2.4 0.827 1.43–3.39*
Head circumference (cm) 43.8 1.3 43.8 1.5 0.865 0.63–0.53*
12 months
Height (cm) 76.9 3.3 75.6 3.1 0.062 0.07–2.62*
≥95 percentile 7 (15.6) 1 (2.1) 0.027 7.5 0.95–58.32
≤5 percentile 2 (4.4) 4 (8.3) 0.678 0.53 0.10–2.77
Weight (kg) 10.466 1.49 9.847 1.26 0.038 0.04–1.20*
≥95 percentile 4 (8.9) 3 (6.3) 0.709 1.42 0.34–6.01
≤5 percentile 3 (6.7) 3 (6.3) 0.630 1.04 0.45–2.38
Weight-for-length > 10% 9 (20.0) 7 (14.6) 0.587 1.37 0.56–3.37
Weight-for-length > 20% 3 (6.7) 1 (2.1) 0.351 3.20 0.35–29.7
PI (kg/m³) 23.0 2.5 22.8 2.0 0.617 0.70–1.18*
Head circumference (cm) 46.9 1.6 46.8 1.7 0.979 0.65–0.67*
18 months
Height (cm) 83.9 3.6 82.2 3.1 0.023 0.22–3.02*
≥95 percentile 6 (13.3) 2 (4.2) 0.150 3.2 0.68–15.04
≤5 percentile 1 (2.2) 6 (12.5) 0.112 0.18 0.02–1.42
Weight (kg) 12.051 1.87 11.318 1.45 0.040 0.04–1.43*
≥95 percentile 5 (11.1) 1 (2.1) 0.104 5.33 0.65–43.91
≤5 percentile 3 (6.7) 4 (8.3) 1.00 0.80 0.19–3.38
Weight-for-length > 10% 10 (22.2) 6 (12.5) 0.275 1.78 0.70–4.49
Weight-for-length > 20% 1 (2.2) 3 (6.3) 0.618 0.36 0.04–3.30
PI (kg/m³) 20.4 2.1 20.3 1.7 0.895 0.75–0.86*
Head circumference (cm) 48.3 1.5 48.4 1.7 0.856 0.76–0.64*
Data are means SDs (Student’s t–test) or n (%) (Fisher’s exact test).
*95% CI of the difference.
[Correction added on 3 December 2014, after first online publication: the statistical data in Table 2 were miscalculated; and these have been
corrected.]
was included in the model. In the multivariate regression offspring of mothers treated with metformin only and off-
analysis, the strongest factor associated with weight of the spring of mothers treated with metformin plus supplemen-
children at the age of 18 months was maternal pre-preg- tal insulin or with insulin alone.’ was irrelevant to the
nancy BMI (P = 0.001, 95% CI 0.04–0.15), but metformin results section; and it has been removed.]
treatment also predicted a higher weight at this time point
(P = 0.036, 95% CI 1.37 to 0.05). Maternal BMI was the
Discussion
only risk factor predicting a child being overweight
(P = 0.013, OR 1.13, 95% CI 1.03–1.25) or obese Main findings
(P = 0.014, OR 1.27, 95% CI 1.05–1.55) at the age of In this randomised study we found that the children exposed
18 months. to metformin prenatally were heavier at 12 months measure-
[Correction added on 3 December 2014, after first online ments and both taller and heavier at 18 months measure-
publication: The sentence, ‘There were no differences in the ments than those exposed to insulin; however their body
growth or development in the subgroup analysis comparing composition defined by PI did not differ. In addition, the
Data are means SDs (Student’s t–test) or n (%) (Fisher’s exact test).
*Without support.
**No meaningful words.
[Correction added on 3 December 2014, after first online publication: the statistical data in Table 3 were miscalculated; and these have been
corrected.]
mean weight-for-length and the proportion of overweight of the study subjects, but they were not otherwise in touch
and obese children did not differ significantly between the with our study.
groups at 6, 12, and 18 months of age. Prenatal exposure to
metformin did not seem to affect the child’s motor, linguis- Interpretation
tic, or social development, assessed at 18 months of age, According to our knowledge, only one randomised study
compared with exposure to insulin. has prospectively evaluated the growth of the offspring of
mothers with GDM treated with metformin or insulin. In
Strengths and limitations the MiG TOFU study, at the age of 2 years, there was no
Our study is the second randomised study and the first difference in the height or in the weight of children exposed
European study comparing metformin and insulin treat- either to metformin or to insulin in utero.14 The study
ments in GDM. In addition to growth, we have assessed reported larger subscapular and biceps skin folds in children
the motor, social, and linguistic development of the infants exposed to metformin, whereas the total fat mass of the
to the age of 18 months. The response rate in our study children did not differ between the groups. The authors
was high, as the questionnaire was returned by 96% of the interpreted this finding as a more favourable pattern of
women who participated in our primary study. growth later in life. The study did not report weight-for--
The limitation of our study is the considerably small sam- length measures.14 Our study found that the children
ple size. As a result, we are not able to detect differences in exposed to metformin were significantly heavier at the age
rare conditions, such as developmental delays, between the of 12 months and both taller and heavier at the age of 18
groups. In the original study, power calculations were made months than children exposed to insulin. In the logistic
to detect a 30% difference in the incidence of macrosomy. regression metformin treatment was a predictor for a higher
Power calculations were not performed in the present study. weight at the age of 18 months. We had no opportunity to
Although the sample size is small, the results of the primary analyse the detailed body composition of the offspring by
study are comparable with other randomised studies and Dual X-ray absorptiometry (DEXA) or other radiological
they may be considered reliable. methods; instead, we used PI to assess the body composi-
We were unable to gather data on the total duration of tion of these infants. Neither mean PI nor the proportion of
breastfeeding, known to affect the weight of the infants. overweight or obese children differed significantly between
We did not have the opportunity to perform check-ups on the groups. There was a tendency to a higher rate of over-
the infants ourselves; however, the Finnish national child weight at the age of 18 months in the metformin group,
welfare clinics use standardised follow-up techniques and but the difference did not reach statistical significance,
calibrated equipment to assess the growth and neurological which may be linked to the small study population.
development of children, and therefore the results can be In a study of 199 children of mothers with polycystic
considered reliable. The nurses and general practitioners ovary syndrome (PCOS), and exposed to metformin or
evaluating these children were not blinded to the allocation placebo in utero, at the age of 1 year the children exposed
to metformin were found to be significantly heavier than min exposure was associated with impaired glucose toler-
those in the placebo group.17 Unfortunately, the study did ance and elevated fasting glucose during a high-fat diet.26
not report the height of the infants.17 The results drawn The importance of environmental factors on growth and
from the studies concerning women with PCOS and treated metabolism should not be ignored, however. Further stud-
with metformin should be interpreted with caution when ies are needed to investigate the body composition and
comparing them with those obtained from being treated insulin sensitivity of the offspring exposed to either metfor-
for GDM. In PCOS the treatment has usually been initiated min or insulin prenatally.
before pregnancy and has often continued throughout the
pregnancy. The intrauterine metabolic environment may be Conclusion
completely different in women being treated for GDM
compared with women being treated for PCOS. Our study found no adverse influence of prenatal metformin
In a recently published study, both birthweight and exposure on motor, linguistic, or social development of the
infant weight, as well as the BMI of children aged over offspring during the first 18 months of life, as compared
2 years, were strongly associated with overweight in later with insulin exposure. The relevance of the finding that
childhood and adolescence.18 In light of this evidence we infants exposed to metformin were heavier at the age of 12
cannot ignore the parallel results of our study and that months and both taller and heavier at 18 months needs to be
conducted by Carlsen et al., indicating that prenatal met- settled in long-term studies.
formin exposure may have an impact on the future growth [Correction added on 3 December 2014 2014, after first
and risks of the offspring.17 online publication: To correct the statistical analysis, the
Maternal metabolism has an effect on fetal programming initial statement ‘children exposed to metformin were hea-
in utero and may cause long-term metabolic consequences vier and taller at the age of 12 and 18 months than chil-
in the offspring by epigenetic mechanisms.19 Both maternal dren exposed to insulin’ has been changed to ‘children
obesity and hyperglycaemia are reported to associate with exposed to metformin were heavier at the age of 12 months
metabolic disturbance and obesity of the offspring, but the and both taller and heavier at 18 months than children
role of pharmacological treatment of GDM in the future exposed to insulin’ throughout the article.]
growth and metabolism of the child has not been investi-
gated separately.20–22 We do not yet know if the glycaemic Disclosure of interests
control during pregnancy has more influence on the subse- There are no conflicts of interest to disclose.
quent growth of the offspring than the mode of treatment.
According to the data from the MiG study, treatment type Contribution to authorship
is not an important variable in relation to neonatal out- HI and MV initiated and designed the study, collected the
comes, compared with glycaemia.23 study population, interpreted the data, and wrote the arti-
A cohort study of 126 children born to women with cle. RK collected a part of the study population. TS
PCOS receiving metformin treatment, begun before designed the questionnaire and participated in the writing
fertilization, and continued throughout pregnancy to of the article. TR initiated and designed the study, collected
decrease pregnancy complications, found no adverse the study population, interpreted the data, and is responsi-
effects on the motor or social development of the chil- ble for the final version of the article.
dren, assessed at the ages of 3, 6, 9, 12 and 18 months
and compared with national gender-spesific data.24 These Details of ethics approval
findings are in line with our study: prenatal metformin Ethics approval for the study was granted by the Ethics
exposion seemed not to have adverse effects on children’s Committee of Northern Ostrobothnia Hospital District on
motor, social or linquistic development.24 20 June 2005 (66/2005).
Metformin reduces insulin resistance and is associated
with beneficial fat redistribution from visceral depots to Funding
subcutaneous deposits in adults.25 Rowan et al. suggested Funding was provided by the Finnish Diabetes Association
that fetuses exposed to metformin in utero might have and the Finnish Medical Society Duodecim (Oulu Divi-
improved insulin action, which may also lead to more ben- sion).
eficial fat distribution and insulin sensitivity later in life.14
On the contrary, in a recent animal experiment on mice, Acknowledgement
prenatal metformin exposure was associated with increased We thank Risto Bloigu (Medical Informatics Group, Uni-
body weight gain, a significant reduction in total body versity of Oulu) for his skilled assistance in univariate and
water content, and an increase in mesenteric fat during a logistic regression analyses. &
high-fat diet phase. In the male offspring prenatal metfor-