Rates of intra-uterine growth, particularly whether restrictive or excessive, can have profound and lasting implications for later adult health( Reference Barker, Gluckman and Godfrey 1 , Reference Barker, Osmond and Simmonds 2 ). In particular, the large-for-gestational age fetus is predisposed to a variety of adverse obstetric and neonatal outcomes( Reference Boulet, Alexander and Salihu 3 , Reference Gregory, Henry and Ramicone 4 ) and in the long term, infants that are at the highest end of the distribution for weight or BMI are more likely to be obese in childhood, adolescence and early adulthood than other infants( Reference Boney, Verma and Tucker 5 , Reference Hermann, Dallas and Haskell 6 ). Glucose is the main fuel for intra-uterine growth( Reference Clapp 7 ). Carbohydrates are classified according to their induced glycaemic response as either high or low glycaemic index (GI)( Reference Jenkins, Wolever and Taylor 8 ). In recent years, interest has been increasing in the role of the GI in pregnancy, and in particular its potential to modulate fetal growth( Reference Louie, Brand-Miller and Markovic 9 , Reference Moses, Barker and Winter 10 ).
We have recently reported the results of the ROLO study, a randomised control trial of a low-GI diet in pregnancy to prevent the recurrence of fetal macrosomia( Reference Walsh, McGowan and Mahony 11 ). It was the first low-GI dietary intervention study to be carried out in pregnant women who previously delivered a macrosomic infant, and the largest randomised control trial of a low-GI diet in pregnancy.
Approximately 50 % of all women in the ROLO study went on to have a macrosomic infant in their second pregnancy, and there was no significant difference in mean infant birth weight between the groups (our primary outcome).
We did find that an introduction of a low-GI diet in pregnancy significantly reduced gestational weight gain (GWG) (our secondary outcome). By 40 weeks of gestation, women in the dietary intervention arm had gained 1·5 kg less than those who received no dietary intervention. A significant proportion of women in each arm of the study exceeded the GWG guidelines, as outlined by the Institute of Medicine (IOM)( Reference Rasmussen and Yaktine 12 ). However, women who received the low-GI diet were significantly less likely to exceed these GWG guidelines, with just 37·7 % doing so compared with 47·9 % of women in the control arm. Our intervention did not include specific weight gain advice for women in the intervention group.
Additionally, we found that the use of a low-GI dietary intervention in pregnancy reduced the incidence of maternal glucose intolerance at 28 weeks of gestation. A significantly higher proportion of women in the control arm at 28 weeks had either a fasting glucose concentration of >5·1 mmol/l or a glucose challenge test result of >7·8 mmol/l.
These findings imply that a once-off dietary education session in pregnancy is beneficial in terms of maternal outcomes, though not in terms of fetal or neonatal outcomes. The ROLO study population comprises a large, well-characterised cohort with uniform, standardised and longitudinal assessments of maternal and fetal size, as well as maternal and fetal insulin resistance and metabolism. Further interrogation of this cohort of pregnant women may not only reveal why some women did respond to the intervention and others did not, but also allow for the characterisation of a particular maternal phenotype that is likely to find this form of intervention particularly useful, and as such, may guide future research in this important area.
The primary objective of the present study was to identify which women, in terms of phenotypic and biochemical characteristics, are most likely to respond to a low-GI dietary intervention in pregnancy to reduce the recurrence of fetal macrosomia, in an attempt to identify from early pregnancy those who might benefit.
Its secondary objective was to identify which characteristics are associated with improvements in maternal glucose homeostasis and GWG.
Experimental methods
The present study is a secondary analysis of all women from the intervention arm of the ROLO study, a randomised control trial. A detailed methodology of the ROLO study has been published previously( Reference Walsh, McGowan and Mahony 11 , Reference Walsh, Mahony and Foley 13 ).
Patient selection
In brief, women were recruited at their first antenatal consultation at 13·8 (sd 2·4) weeks. Subjects were excluded if they had any underlying medical conditions, were < 18 years of age, had previous gestational or pre-existing type 1 or type 2 diabetes or were unable to give full informed consent.
Ethical approval
Ethical approval was granted by the National Maternity Hospital Ethics Committee (2007).
Data collection and trial management
At their first antenatal visit, all patients had their weight and height recorded and their BMI calculated. Concentrations of fasting blood glucose, insulin and leptin were measured and a mid-upper arm circumference recorded. Questions relating to maternal lifestyle habits, including physical activity, smoking, breast-feeding practices and education level, were included in the FFQ and given to the subjects at their first antenatal visit. Maternal education level was used as a surrogate marker of socio-economic status. Education level was categorised as one of six categories: 6, complete third level (higher-level degree); 5, some third level (certificate/diploma); 4, complete second level; 3, some second level; 2, primary education only; 1, no education.
Maternal weight was recorded at each antenatal consultation and total GWG calculated according to the IOM guidelines( Reference Rasmussen and Yaktine 12 ). At 28 weeks of gestation, repeat fasting blood glucose, insulin and leptin concentrations were measured and glucose challenge testing 1 h after a 50 g glucose load was carried out. At delivery, infant birth weight, length and head circumferences were recorded, and a cord blood sample for analysis of fetal glucose, leptin and C-peptide concentrations was taken. The present controlled trial was registered as ISRCTN54392969.
Dietary intervention
The intervention comprised one dietary education session lasting 2 h in groups of two to six women with a research dietitian. The mean gestational age of those attending the dietary session was 15·7 (sd 3·0) weeks.
The education session focused on the GI. Women were encouraged to choose as many low-GI foods as possible and to exchange high-GI carbohydrates for low-GI alternatives. Women received written resources about low-GI foods after the education session. The recommended low-GI diet was euenergetic, and women were not advised to reduce their total energy intake.
Dietary assessment
All subjects completed a 3 d food diary during each trimester of pregnancy where the type and amount of all foods and beverages consumed were recorded over three consecutive days. Subjects were encouraged to include 1 weekend day during the recording period. Subjects were instructed to quantify their food consumed using either the manufacturer's weight on the food packaging or standard household measures. Dietary data were entered into WISP version 3.0 (Tinuviel Software). The 3 d food diary was used to assess the dietary GI of each patient and the response to the dietary intervention.
Laboratory methods
Multianalyte profiling was performed on the Luminex Magpix system (Luminex Corporation). Plasma concentrations of leptin, insulin and C-peptide were determined by the Human Endocrine Panel. Maternal insulin resistance was calculated using the homeostasis model assessment (HOMA) index( Reference Matthews, Hosker and Rudenski 14 ):
$$\begin{eqnarray} HOMA\ score = (fasting\ insulin\ (\mu U/ml)\times fasting\ glucose\ (mmol/l))/22.5. \end{eqnarray}$$
Fetal insulin resistance was assessed by cord blood C-peptide estimation.
Statistical analysis
Maternal demographic characteristics in early pregnancy, details from the previous pregnancy notes and maternal glucose and insulin assessments were compared using Student's t test, Mann–Whitney U test and χ2 test for the following groups:
-
(1) Those who did and those who did not have a recurrence of fetal macrosomia.
-
(2) Those who did and those who did not exceed the GWG guidelines as outlined by the IOM.
-
(3) Those who did and those who did not have impaired glucose tolerance in later pregnancy.
Recurrence of fetal macrosomia was assessed as an infant birth weight of ≥ 4000 g in the second pregnancy. Impaired glucose tolerance was diagnosed when women had either a fasting glucose concentration of ≥ 5·1 mmol/l or a glucose challenge test of >7·8 mmol/l at 28 weeks of gestation.
Results
A total of 372 women in early pregnancy from the intervention arm of the ROLO study were included in the secondary analysis. Compared with the control group, maternal GI was significantly reduced in the intervention group at the second and third trimesters (56·1 (sd 4) v. 57·8 (sd 4), P< 0·001 and 56 (sd 3·7) v. 57·7 (sd 3·9), P< 0·001, respectively).
Comparison of the maternal characteristics of those who did and those who did not have a recurrence of fetal macrosomia following the low-glycaemic index dietary advice in pregnancy
Overall, fetal macrosomia (birth weight ≥ 4000 g) recurred in 50·8 % (n 189/372) of women who received the low-GI dietary advice in pregnancy. A comparison of the maternal characteristics of those 189 women and 183 women who delivered their second infants weighing < 4000 g is outlined in Table 1.
Table 1 Comparison of the maternal characteristics of those who did and those who did not have a recurrence of fetal macrosomia following the low-glycaemic index dietary advice in pregnancy* (Mean values and standard deviations for normally distributed data; median values and interquartile ranges for non-parametric data (homeostasis model assessment (HOMA) index and leptin concentration))
* Independent Student's t test used for parametric data and Mann–Whitney U test used for non-parametric data.
† HOMA = (fasting insulin (μU/ml) × fasting glucose (mmol/l))/22·5.
‡ Education level was categorised as one of six categories: 6, complete third level (higher-level degree); 5, some third level (certificate/diploma); 4, complete second level; 3, some second level; 2, primary education only; 1, no education.
During the first pregnancy, no difference was observed between the two groups in terms of first birth weight or GWG.
In the second pregnancy, there was no difference observed between the two groups in terms of mean maternal BMI in early pregnancy (27·12 (sd 5·2) v. 26·36 (sd 5·1) kg/m2, P= 0·1); however, women who went on to deliver a second macrosomic infant were found to be heavier (75·5 (sd 14·0) v. 71·8 (sd 14·8) kg, P= 0·02) and taller (167·1 (sd 6·1) v. 165·2 (sd 6·7) cm, P= 0·01). They also had higher mid-upper arm circumferences (29·9 (sd 3·8) v. 29·1 (sd 4·1) cm, P= 0·04), a marker of maternal adiposity.
No difference was observed between those who did and those who did not go on to deliver a second macrosomic infant in terms of maternal fasting glucose concentrations or insulin resistance (HOMA) in early pregnancy, though a significant difference in leptin concentrations was observed during early pregnancy (Table 1).
Interestingly, maternal education level did differ between the two groups, with a higher mean education level in those who went on to deliver a second macrosomic infant (4·4 (sd 2·1) v. 3·9 (sd 2·4), P= 0·03).
Comparison of the maternal characteristics of those who did and those who did not exceed the gestational weight gain guidelines as outlined by the Institute of Medicine following the low-glycaemic index dietary advice in pregnancy
A comparison of the maternal characteristics of those who did and those who did not exceed the GWG guidelines as outlined by the IOM is contained in Table 2.
Table 2 Comparison of the maternal characteristics of those who did and those who did not exceed the gestational weight gain (GWG) guidelines following the low-glycaemic index dietary advice in pregnancy* (Mean values and standard deviations for normally distributed data; median values and interquartile ranges for non-parametric data (homeostasis model assessment (HOMA) index and leptin concentration))
* Independent Student's t test used for parametric data and Mann–Whitney U test used for non-parametric data.
† HOMA = (fasting insulin (μU/ml) × fasting glucose (mmol/l))/22·5.
‡ Education level was categorised as one of six categories: 6, complete third level (higher-level degree); 5, some third level (certificate/diploma); 4, complete second level; 3, some second level; 2, primary education only; 1, no education.
Women who exceeded the GWG guidelines in their second pregnancies following the low-GI dietary intervention were heavier at the beginning of their first pregnancies (76·01 (sd 13·8) v. 69·8 (sd 13·2) kg, P= 0·001), and also had higher GWG during that pregnancy (17·5 (sd 4·5) v. 12·9 (sd 4·8) kg, P< 0·001).
Their early pregnancy BMI were higher in the index pregnancy (28·3 (sd 5·1) v. 26·2 (sd 5·4) kg/m2, P< 0·01), with no difference being observed in maternal height. They also had more adiposity with larger mid-upper arm circumferences (30·5 (sd 3·8) v. 29·2 (sd 3·7) cm, P= 0·005). There was no difference observed between those who did and those who did not exceed the GWG guidelines following the low-GI dietary intervention in terms of maternal fasting glucose concentrations or insulin resistance (HOMA) in early pregnancy, though serum leptin concentrations were found to be higher in those who exceeded the GWG guidelines during early pregnancy (Table 2).
Comparison of the maternal characteristics of those who did and those who did not have impaired glucose tolerance following the low-glycaemic index dietary advice in pregnancy
A comparison of the maternal characteristics of those who did and those who did not have impaired glucose tolerance following the low-GI dietary intervention in pregnancy is outlined in Table 3.
Table 3 Comparison of the maternal characteristics of those who did and those who did not have impaired glucose tolerance following the low-glycaemic index dietary advice in pregnancy* (Mean values and standard deviations for normally distributed data; median values and interquartile ranges for non-parametric data (homeostasis model assessment (HOMA) index and leptin concentration))
* Independent Student's t test used for parametric data and Mann–Whitney U used for non-parametric data.
† HOMA = (fasting insulin (μU/ml) × fasting glucose (mmol/l))/22·5.
‡ Education level was categorised as one of six categories: 6, complete third level (higher-level degree); 5, some third level (certificate/diploma); 4, complete second level; 3, some second level; 2, primary education only; 1, no education.
At 28 weeks of gestation, women who had either a fasting glucose concentration of >5·1 mmol/l or a glucose challenge test result of >7·8 mmol/l had higher BMI in early pregnancy than those who did not (27·9 (sd 4·9) v. 26·1 (sd 4·5) kg/m2, P= 0·005). However, there was no difference in BMI between the two groups in their first pregnancies (25·4 (sd 4·1) v. 25·6 (sd 4·5) kg/m2, P= 0·7), with those who developed glucose intolerance having significantly increasing BMI between pregnancies (25·4 (sd 4·1) v. 27·9 (sd 4·9) kg/m2, P= 0·008).
They also had significantly higher fasting glucose concentrations in early pregnancy (4·51 (sd 0·3) v. 4·42 (sd 0·3), P= 0·05) with more insulin resistance and higher leptin concentrations than those who did not go on to develop glucose intolerance in later pregnancy. However, there was no difference in education level between those who did and those who did not develop glucose intolerance following the low-GI dietary intervention in pregnancy.
Discussion
The present secondary analysis of a cohort of pregnant women from the intervention arm of the ROLO study has defined the maternal characteristics in early pregnancy associated with a response to a low-GI dietary intervention in pregnancy with respect to three important parameters of obstetric risk: infant birth weight; maternal glucose metabolism; GWG.
We found that women who went on to deliver a second macrosomic infant were taller with a higher mean education level. Women who exceeded the GWG guidelines were found to be heavier in both their first and second pregnancies, though no difference was observed in maternal height. They also had more adiposity with larger mid-upper arm circumferences. At 28 weeks of gestation, women who developed glucose intolerance had a higher BMI with significantly higher fasting glucose concentrations in early pregnancy with more insulin resistance. For each analysis (birth weight, weight gain and glucose intolerance), women who responded to the low-GI dietary intervention had lower fasting leptin concentrations in early pregnancy than those who did not.
Primary analysis of this cohort demonstrated no difference in the recurrence of fetal macrosomia following the commencement of a low-GI diet in pregnancy. However, the diet was associated with a significant improvement in maternal outcomes, with less GWG and less glucose intolerance.
Interestingly, the findings of the present secondary analysis revealed a similar disparity between maternal and fetal outcomes. There was no difference between those who did and those who did not go on to deliver a second macrosomic infant in terms of maternal BMI in early pregnancy, though mothers who were taller were more likely to deliver a second infant weighing >4000 g. In contrast, maternal BMI was significantly higher in women who went on to exceed the GWG guidelines as outlined by the IOM and in those who developed glucose intolerance at 28 weeks of gestation. Similarly, there was no difference between those who did and those who did not go on to deliver a second macrosomic infant in terms of maternal fasting glucose concentrations, insulin resistance (HOMA indices) or leptin concentrations in early pregnancy; however, those who developed glucose intolerance had significantly higher fasting glucose concentrations in early pregnancy with more insulin resistance (higher HOMA indices) and higher leptin concentrations than those who did not.
These findings confirm that the maternal metabolic environment in early pregnancy is important in determining later risks of excessive weight gain and metabolic disturbance. In contrast, infant birth weight in this healthy, non-diabetic population is mediated to a greater extent by genetic factors, such as maternal height.
Notably, maternal leptin concentration in early pregnancy was the single marker that differed consistently in each of the three analyses of the present study. Despite low-GI dietary advice, women who gave birth to second macrosomic infants, exceeded the GWG guidelines or developed glucose intolerance in later pregnancy all had significantly higher leptin concentrations in early pregnancy than those who did not. There raises an interesting possibility that perhaps maternal leptin concentration is a more sensitive marker than maternal BMI, adiposity or, indeed, insulin resistance of an underlying predisposition towards pregnancy complications in such a cohort of women.
The present findings in relation to maternal education level, a surrogate marker of socio-economic status, are interesting. Again, we observed a maternal/fetal disparity, with education level being associated with the recurrence risk of fetal macrosomia, but not with maternal weight gain or glucose homeostasis. These findings would suggest that women with lower education levels are perhaps more likely to respond to dietary education in pregnancy. Perhaps this group is more susceptible to aberrant fetal growth secondary to poor dietary choices, and should be considered as a group likely to respond to simple dietary education measures. These findings highlight the public health importance of education during pregnancy about healthy lifestyle choices.
Similarly, our findings in relation to the effect of interval pregnancy weight gain and glucose intolerance are important for public health education. We found that women who developed glucose intolerance in their second pregnancies had significantly increased their BMI from the first to second pregnancy (from a mean of 25 kg/m2 to just under 29 kg/m2). This highlights the important role of post-partum weight retention and, indeed, weight gain in dictating future metabolic risk, for both the mother and the baby. It would suggest that obstetricians and carers should advise all women, including those with a normal BMI, of the potential hazards of interval pregnancy weight gain for future pregnancies.
There are a number of potential limitations to these data that warrant further consideration. The present study is a subgroup analysis and therefore likely to be underpowered to draw concrete conclusions. This analysis was performed with the intention of further interrogation of the ROLO study data to further our understanding of why some women did and some did not respond to the dietary intervention; as such, multiple comparisons were made which raised the possibility of type 1 statistical error.
In addition, infant birth weight alone is a relatively crude, if clinically applicable, assessment of intra-uterine growth. Follow-up assessment of neonatal and childhood adiposity, currently in progress, will add invaluable data to separate physiological from pathological fetal growth.
These findings have important implications for clinical practice. They suggest a particular maternal phenotype that is likely to respond, or not, to low-GI dietary intervention in pregnancy. Importantly, we identified no harm associated with the introduction of a low-GI diet in pregnancy. These findings may allow for tailoring of dietary intervention based on maternal characteristics in early pregnancy, such as early pregnancy weight and glucose, in order to improve outcomes in all maternal phenotypes. Perhaps a more intensive regimen, with more regular feedback and re-enforcement, may be necessary for those who fit certain criteria at first antenatal consultation. A recent publication by Rhodes et al. ( Reference Rhodes, Pawlak and Takoudes 15 ) reported the results of a pilot study comparing a low-glycaemic load diet with a low-fat diet in a group of forty-six overweight and obese pregnant women. These authors employed a much more intensive regimen including weekly reinforcement and home food delivery. Their results in this particular patient group reflect our experience with little or no effect on infant birth weight but an improvement in maternal outcomes. Perhaps a once-off dietary education session could be considered for all women at risk of excessive gestation weight gain or impaired glucose tolerance, with a more intensive regimen employed for overweight and obese women in early pregnancy.
The benefits of an intervention that reduces maternal GWG, irrespective of maternal BMI, cannot be underestimated. Maternal weight gain during pregnancy has been reported to be independently linked to adverse obstetric outcomes for all BMI categories( Reference Boulet, Alexander and Salihu 3 , Reference Gregory, Henry and Ramicone 4 ). There are also potential maternal implications to excessive GWG, such as an increased operative delivery rate, a higher likelihood of post-partum weight retention and a predisposition to later obesity( Reference Scholl, Hediger and Schall 16 , Reference Rooney and Schauberger 17 ). Similarly, there is a clear association between maternal hyperglycaemic, even at levels below those diagnostic of gestational diabetes, and a variety of adverse maternal, fetal and neonatal outcomes( Reference Metzger and Lowe 18 ).
Further work is still necessary to determine whether pregnancy interventions can reduce the recurrence of fetal macrosomia in euglycaemic women. It is clear from our work that a low-GI diet alone is not sufficient to combat the problem, despite conferring maternal benefits.
The present secondary analysis has confirmed that infant birth weight is a complex interplay between genetic and environmental factors. There was no difference between those who did and those who did not go on to deliver a second macrosomic infant in terms of maternal BMI in early pregnancy, though mothers who were taller with higher education levels were more likely to deliver a second infant weighing >4000 g. There was also no difference between those who did and those who did not go on to deliver a second macrosomic infant in terms of maternal fasting glucose concentrations, insulin resistance or leptin concentrations in early pregnancy. This would suggest that in this cohort of women, at least, genetic factors played a more significant role in the recurrence risk of fetal macrosomia than did the maternal environment and metabolic milieu. This may be at least in part explained by the strict inclusion criteria of the ROLO study. This group of study mothers was specifically selected to allow us to examine the effect of a low-GI diet among healthy euglycaemic women. Perhaps our selection criteria introduced a degree of selection bias towards women who are giving birth to larger babies due to genetic potential rather than due to aberrations in maternal metabolism.
In conclusion, there are no adverse outcomes associated with the use of a low-GI diet in pregnancy. There are potential benefits, in particular in terms of limiting maternal GWG to within the IOM guidelines, and an improvement in maternal glucose homeostasis. These findings were identified following a single, formal small-group dietetic session in early pregnancy in the ROLO study. This suggests that this type of simple dietary intervention is adequate in improving maternal nutrition in a general population of women.
The present analysis would suggest that modifications of this pregnancy intervention may be necessary in order to optimise outcomes for women at the higher end of the spectrum in terms of weight, adiposity and insulin resistance in order to improve these important benefits for all.
Acknowledgements
The authors thank the staff at National Maternity Hospital and the mothers who participated in the study.
The present study was funded by the Health Research Board: Health Research Centre for Diet, Nutrition and Diabetes, with additional financial support from the National Maternity Hospital Medical Fund. None of the funding sources had a role in the trial design or manuscript preparation.
The authors' contributions were as follows: F. M. M. conceived and designed the study; G. C. assisted with the data collection; R. M. M. and M. E. F. contributed to the study design, data collection and manuscript preparation; J. M. W. carried out the analysis and wrote the manuscript. All the authors reviewed and revised the final version of the manuscript.
None of the authors has any conflict of interest to declare.
