Maternal low glycaemic index diet, fat intake and postprandial glucose influences neonatal adiposity

12 Pages·2014·0.34 MB·English

Horanetal.NutritionJournal2014,13:78
http://www.nutritionj.com/content/13/1/78
RESEARCH Open Access
Maternal low glycaemic index diet, fat intake and
postprandial glucose influences neonatal
–
adiposity secondary analysis from the ROLO
study
Mary K Horan1, Ciara A McGowan1, Eileen R Gibney2, Jean M Donnelly1 and Fionnuala M McAuliffe1*
Abstract
Background: The inutero environment is known to affect fetal development however many of the mechanisms
bywhich this occurs remain unknown.The aim ofthis study was to examinethe association between maternal
dietary macronutrientintake and lifestyle throughout pregnancy and neonatal weightand adiposity.
Methods: This was ananalysis of542 mother and infant pairs from theROLO study (Randomised cOntrol trial of
LOw glycaemic index diet versus no dietary interventionto prevent recurrence of fetalmacrosomia). Food diaries as
well as food frequency and lifestyle and physical activity questionnaires were completedduring pregnancy.
Maternal anthropometry was measured throughout pregnancy and neonatal anthropometry was measured atbirth.
Results: Multiple linear regression analysis revealedthe main maternal factor associated with increased birth weight
was greater gestational weight gain R2 23.3% (F=11.547,p<0.001). The main maternal factor associated with
adj
increasedbirthlengthwasnon-smokingstatusR2 27.8%(F=6.193,p<0.001).Neonatalcentraladiposity(determined
adj
usingwaist:lengthratio)wasnegativelyassociatedwithmaternalage,andpositivelyassociatedwiththefollowing
parameters:smokingstatus,maternalpre-pregnancyarmcircumference,percentageenergyfromsaturatedfatin
latepregnancy,postprandialglucoseat28weeksgestationandmembershipofthecontrolgroupwithapositive
trendtowardsassociationwithtrimester2glycaemicloadR2 38.1%(F=8.000,p<0.001).
adj
Conclusions:Severalmaternaldietandlifestylefactorswereassociatedwithneonatalanthropometry.Low
glycaemicindexdietaryinterventioninpregnancywasfoundtohaveabeneficialeffectonneonatalcentral
adiposity.Additionally,centraladipositywaspositivelyassociatedwithmaternaldietaryfatintakeandpostprandial
glucosehighlightingtheimportantroleofhealthydietinpregnancyinpromotingnormalneonataladiposity.
Trialregistration:CurrentControlledTrialsISRCTN54392969.
Keywords:Maternaldiet,Pregnancy,Neonatalbodycomposition,Neonataladiposity,Lowglycaemicindexdiet
Background fetus differently depending on the stage of pregnancy
The in utero environment has been found to affect fetal [2,4,5]. Thematernaldietduringpregnancyisparticularly
development in a variety of ways from cognitive develop- important as the mother is relied on to provide all of the
ment[1]todevelopment ofthefetal organs[2]togrowth nutrients required for the fetus to grow and develop [6],
andfatdeposition[3].Environmentalfactorstowhichthe however maternal pre-pregnancy nutritional status and
pregnant woman is exposed result in epigenetic changes lifestyle have also been found to be important [7]. Pre-
which impact on fetal genetic transcription and affect the pregnancyoverweightandobesityaswellasexcessiveges-
tational weight gain have been found to result in fetal
macrosomia i.e. a birthweight of ≥4 kg [7]. Macrosomic
*Correspondence:[email protected]
1UCDObstetricsandGynaecology,SchoolofMedicineandMedicalScience, infants are at increased risk of developing metabolic syn-
UniversityCollegeDublin,NationalMaternityHospital,Dublin2,Ireland drome in later life and this effect may persist in later
Fulllistofauthorinformationisavailableattheendofthearticle
©2014Horanetal.;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreative
CommonsAttributionLicense(http://creativecommons.org/licenses/by/4.0),whichpermitsunrestricteduse,distribution,and
reproductioninanymedium,providedtheoriginalworkisproperlycredited.TheCreativeCommonsPublicDomain
Dedicationwaiver(http://creativecommons.org/publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle,
unlessotherwisestated.
Horanetal.NutritionJournal2014,13:78 Page2of12
http://www.nutritionj.com/content/13/1/78
generations through intergenerational programming [8]. controltrialof800secundigravidawomenwithaprevious
Macrosomiaalsoincreasesobstetricriskssuchasshoulder macrosomic baby (>4 kg) randomised to receive low gly-
dystocia, maternal anal sphincter injury, instrumental va- caemic index (GI) dietary advice versus usual care (no
ginaldeliveryandemergencycaesareansection[9]. dietary advice) to reduce recurrence of macrosomia. De-
The hyperglycaemia-hyperinsulinism theory proposes tailedmethodologyand resultsoftheROLOstudy,which
that the mother’s intake of carbohydrate and her natural was carried out in the National Maternity Hospital,
pregnancy-related progressive insulininsensitivity results Ireland, have previously been published [25,26] but in
in higher levels of maternal blood glucose which is brief; the primary outcome was birthweight and the sec-
transferred to the fetus [10]. In response, the fetus pro- ondaryoutcomesweregestationalweightgainandglucose
duces its own insulin which then acts as a growth hor- intolerance.LowGIdietaryadvicewasgivenatweek14of
mone resulting in increased fetal growth and adiposity pregnancy while demographic, well-being and lifestyle
[9,10]. Studies support this hypothesis during gestational questionnaireswerereturnedby28weeksgestation.3-day
diabetes [11] and also within normal limits of maternal food diaries were also completed in each trimester of
blood glucose [12]. As such it is important for maternal pregnancyandusedtodeterminetheglycaemicindexand
blood glucose levels to be maintained within normal glycaemic load of the women’s diets. The control group
levels to ensure fetal glucose levels, and subsequent in- received nodietary advice and had routineantenatal care.
sulinlevelsarealso maintained within normalranges. The ROLO study found that the intervention group sig-
Otherdietary factorsincludingmaternaldietaryenergy nificantly lowered their glycaemic index and glycaemic
andproteinintakehavealsobeenfoundtoinfluenceneo- load and had significantly lower gestational weight gain
natal anthropometry [13-15] while the quality of macro- and glucose intolerance but birthweight or risk of macro-
nutrients also appears to have a role, in particular the somia was not significantly reduced [25,27]. This study
quality of fat i.e. saturated or trans-fat vs polyunsaturated was conducted according to the guidelines laid down in
fatintake[16,17]andthetypeofproteini.e.dairyvsmeat the Declaration of Helsinki with ethics approval from the
protein [18]. Maternal micronutrient intakes also affect National Maternity Hospital review board and written in-
neonatal body composition however even less data is formedmaternalconsent.
available in this area except in the area of deficiency and
supplementation in the developing world in order to re- Inclusionandexclusioncriteria
duce preterm birth and small for gestational age infants Participants were secundigravida women with a previous
[19,20]. Therefore, this study focuses on the association macrosomic baby (>4 kg). They were required to have
between maternal macronutrient intake throughout preg- sufficient literacy and English language fluency to
nancy and neonatal body composition. The relationship understand the intervention and be capable of com-
between neonatal size and maternal dietary intake is not pleting questionnaires. Women were only included if
clear cut since, in addition to diet, a variety of maternal they had healthy, singleton pregnancies with no intra-
characteristics and lifestyle factors including physical ac- uterinegrowthabnormalities.
tivity, socioeconomic and demographic status, stress,
smoking, alcohol consumption and other drug intake as Maternaldemographics,lifestyleandwell-being
wellasgeneticfactorsalsoaffectfetalgrowthanddevelop- Ofthe800participantsoftheROLOstudy,542completed
ment[4,21-23].Duetothecomplexnatureofthedetermi- questionnairesinthefirsthalfofpregnancyexploringvari-
nants of neonatal size and adiposity and the fact that the ous background socioeconomic and socio-demographic,
most commonly reported anthropometric parameters and lifestyle variables. Questions from SLAN (Survey of
“weight”and“length”areverylimitedmeasuresofadipos- Lifestyle, Attitudes and Nutrition in Ireland) [28] relating
itywhichgivenoinformationonbodyfatdistribution[24] to lifestyle habits were completed, including questions on
thereremainsapaucityofdatainthearea.Theaimofthis physical activity, smoking and educational attainment.
study was to use a cohort from the ROLO (Randomised Well-being was measured using the World Health Organ-
cOntroltrialofLOwglycaemicindexdietversusnodietary isation 5-Item Index (WHO-5) expressed as a percentage
intervention to prevent recurrence of fetal macrosomia) scorefrom 0-100% i.e. the lowest to highest possible well-
studytoexaminetheeffectofmaternalbodycomposition, beingscores[29].
demographiccharacteristics,macronutrientintakeandlife-
style both pre-pregnancy and throughout pregnancy on Maternalandneonatalanthropometry
neonatalweightandadiposity. Maternal weight, height and mid-upper arm circumfer-
ence were measured at the first antenatal consultation
Methods and BMI calculated. Maternal weight was also measured
542motherandinfantpairsfromtheROLOstudywerein- at each subsequent consultation and gestational weight
cludedinthisanalysis.TheROLOstudywasarandomised gaincalculated.
Horanetal.NutritionJournal2014,13:78 Page3of12
http://www.nutritionj.com/content/13/1/78
Neonatalweight,length,mid-upperarm,abdominal,hip In brief, all food diaries and food frequency question-
and thigh circumference, and biceps, triceps, subscapular naires were entered by a trained dietitian/clinical nutri-
and thigh skinfold measurements were taken at birth. tionist with the use of the household measures and UK
Weight and length were measured for all 542 neonates Food Standards Agency average portion sizes [38]. Food
while other anthropometric measurements began to be DiarieswereanalysedusingTinuvielWISPsoftware,ver-
taken later in the study and therefore complete data were sion 3.0, in which the food composition tables used are
availablefor266neonates.Waist:hip,waist:lengthandtri- derivedfromthe6theditionofMcCanceandWiddowson’s
cepsskinfold:subscapularskinfold ratiosas wellas sum of Food Composition Tables [39]. Food Frequency Question-
triceps and subscapular skinfold thicknesses and sum of naires were analysed using Tinuviel QBuilder software,
all skinfold thicknesses were calculated as a measure of version 2.0, which also uses food compositions from the
neonataladiposity. 6th editionofMcCanceand Widdowson’sFood Compos-
ition Tables [39]. GI values were updated in WISP and
Maternaldietaryintake QBuilder from 2002 values using the 2008 International
3-day food diaries were completed in each trimester of TablesofGlycaemicIndexValuesandothermorerecently
pregnancy and used to determine macronutrient intake publishedGIvalues[40,41].
as well as the glycaemic index and glycaemic load of the
women’s diet. Macronutrients were expressed as a per- Oralglucosechallengetest(GCT)
centage of total energy. Macronutrientintake duringeach At 28weeksgestation,fastingglucose wasmeasured and
trimesterofpregnancywasexaminedseparately.Underre- aglucose challenge test(GCT) measuring serum glucose
portingwasexaminedusingGoldbergratiosietheratioof one hour post 50 g glucose load was performed. Post-
energyintaketoestimatedbasalmetabolicrate[30].Basal prandial blood glucose levels were recorded with re-
metabolic rate was calculated using Schofield equations sults ≥7.8 mmol/L categorised as glucose intolerance
and a Goldberg ratio of ≤0.9 was used to identify definite [42,43]. GCT is normally carried out as part of institu-
underreporters[30-32]. tional policy only if women have specific risk factors for
Cluster analysis had been previously completed for the gestational diabetes. If women have a blood glucose level
control group of this cohort using 3 day food diaries to of>8.3 mmol/L 1 hour post GCT, formal glucose toler-
define food group intake at each trimester which was ance testing is carried out on a different day to rule out
then analysed using k-means cluster analysis resulting in gestational diabetes. However, all participants of the
the identification of 2 main clusters of “healthy” and ROLO study underwent GCTas part of the study proto-
“unhealthy” individuals with regard to diet [33]. In brief col. They were then referred for glucose tolerance testing
the unhealthy dietary cluster was characterised by sig- only if blood glucose level was>8.3 mmol/L in accord-
nificantlyhigherintakesof white bread, refinedbreakfast ancewithinstitutionalpolicy.
cereals, confectionery, chips, processed meats and high-
energy beverages. The second cluster was characterised Statisticalanalysis
by significantly higher intakes of wholegrain breads and Statistical analysis was completed using SPSS (Statistical
breakfast cereals, fruit, vegetables, fruit juice, fish, low- Package for the Social Sciences) software version 20.0.
fat milk and white meat [33]. Dietary intake over the Statistical analyses involved correlations, independent
past 3 months was examined using the self-administered sample t-tests, ANOVA, ANCOVA, simple and multiple
170 item SLAN (Survey of Lifestyle, Attitudes and Nu- linear regression modelling. The intervention and control
trition in Ireland) food frequency questionnaire (FFQ) groups were analysed both separately and together to en-
which was given in early pregnancy and returned by sure all results were representative of both groups. Be-
28weeksgestation.TheSLANFFQwasoriginallyadapted cause there was no difference in neonatal anthropometry
from the European Prospective Investigation of Cancer except for thigh circumference [44] and waist:length ratio
(EPIC) study and has been validated for use in the Irish betweenthecontrolandinterventiongroups,groupswere
adultpopulation[34,35].Dietarydatafromthe completed not analysed separately for final analysis but group was
FFQs was used to create a Dietary Approaches to Stop controlledforinallfinalmodels.Analysiswasalsocarried
Hypertension (DASH) index i.e. a score measuring the out both including and excluding definite underreporters
levelofconcordancewiththeDASHdietfrom0whichin- ofdietaryintakeandsincetheexclusionoftheseunderre-
dicates total non-concordance to 11 which indicates total porters did not change any of the significant associations
concordance [36]. The DASH diet was developed for the but did decrease the power of the analysis, dietary under-
reduction of hypertension but, as with other dietary indi- reporterswerenotexcludedwhenpresentingfinalmodels
ces,alsoindicatesoverallhealthydiet[37]. with the exception of the waist:length circumference
Full methodology for entry and analysis of the dietary whose associations were affected. In order to determine
intake ofparticipantshaspreviously been published [27]. whether the loss of significance of some associations in
Horanetal.NutritionJournal2014,13:78 Page4of12
http://www.nutritionj.com/content/13/1/78
this model was due to a loss of power or actually due to There was also no difference in physical activity levels
the effect of underreporting, underreporting was then betweenthecontrolandinterventiongroupsatbaseline.
controlledforinthismodelandboththeadjustedandun-
adjusted models presented. Healthiness of the diet was Underreporting
compared to neonatal anthropometry by examining diet- There was no difference in underreporting between the
ary clusters (as defined above) and the DASH dietary intervention and control groups (12.7% vs 9.7% definite
index using simple linear regression. In order to build underreporters respectively, p=0.821). There was no
multipleregressionmodels,variables(includingmacronu- difference in any of the associations in the final models
trient intake for each trimester, parental height, weight when definite underreporters were removed with the ex-
and BMI,gestational weight gain, maternalphysical activ- ception of waist:length ratio, however the power of these
ity, smoking status, ethnicity, age, marital/partner status, models was reduced. Therefore the final models pre-
oral glucose challenge test results, alcohol intake and gly- sented in Table 2. include underreporters. When definite
caemic index and load status) were first analysed using underreporterswereremovedfromthewaist:lengthmodel,
correlations, independent sample t-tests and ANOVA as significant positive associations remained with maternal
appropriate. Variables that were found to be significantly smokingandtrimester3saturatedfatintakewhiletheas-
associatedwithneonatalanthropometrywerefurtherana- sociations with maternal age, mid-upper arm circumfer-
lysedusingsimplelinearregressionthenimputedintothe ence, trimester 2 GL, postprandial glucose and group lost
final multiple regression model for well-being using a significance. In order to determine whether this was due
forced enter and backwards stepwise approach. Variables to a lack of power or to underreporting being an actual
thatwerestatisticallysignificantlyassociatedwithneonatal confounder, underreporting was added to the final mul-
anthropometry using simple linear regression were then tiple linear regression model in order to control for it.
included in a backwards stepwise multiple regression Subsequently, all factors regained statistical significance
block resulting in any non-significant variables being dis- withtheexception oftrimester 2 GLwhich originallyhad
carded from the model in a stepwise manner. Variables only showed a trend toward a significant association, and
knowntoeffectneonatalsize(educationlevelasamarker postprandial glucose which showed a trend (p=0.050)
of socioeconomic status, maternal pre-pregnancy BMI, towards a positive association with waist:length circum-
length of gestation and neonate gender) were controlled ference when the model was adjusted for underreporting.
for using a forced enter multiple regression block in all This adjusted final model is presented in Table 2 in
models. As mentioned, membership of the control or additiontotheunadjustedmodel.
interventiongroupwasalsoincludedbyforcedentermul-
tipleregressioninthesemodels.Multiplelinearregression Maternalandneonatalanthropometry
resulted in a best and final model and those that were There was no difference in neonatal weight, length or
statisticallysignificantoverall(p<0.05)were chosenas other anthropometric measurements between the inter-
those which best predicted neonatal anthropometric vention and control groups except in thigh circumfer-
measurements. ence measurement which has previously been described
[44] and in neonatal waist:length ratio which was signifi-
Results cantly lower in the intervention group. Mean maternal
Demographics,lifestyleandwell-being and neonatal anthropometric measurements are shown
Maternal characteristics are listed in Table 1. Maternal inTable1.Maternal anthropometry wasnot significantly
characteristics did not differ between the control and differentatbaselinebutgestationalweightgainwashigher
intervention groups except for gestational weight gain in the control group than the intervention group as de-
[25,27], glucose intolerance and maternal well-being scribedabove.
score [29] as previously described. 91.4% of the women Associations of maternal characteristics and macronu-
were of “white Irish”, 6.7% of “white other”, 0.3% of trient intakewithneonatal anthropometry observed using
“African”, 0.5% of “Chinese”, 0.1% of “Indian” and 1.0% simple linear regression are listed in Table 3. Statistically
of “Filipino/South East Asian” ethnicity. Again, there significant multiple linear regression models for the asso-
was no difference in ethnicity between the control and ciation between maternal characteristics and neonatal an-
intervention groups (p=0.159). 78.1% of the women thropometryareshowninTable2.Nosignificantmultiple
had achieved 3rd level education while 21.9% had not linear regression models existed for; hip circumference,
and there was no difference in these rates between the triceps skinfold thickness, biceps skinfold thickness, thigh
control and intervention groups (p=0.680), similarly skinfold thickness, waist:hip circumference ratio, sum
there was no difference between reported smoking status of subscapular and triceps skinfold thickness or for sum of
between women in the control and intervention groups all skinfold thicknesses. Statistically significant multiple re-
with 4.0% smokers and 96.0% non-smokers (p=0.208). gressionmodelsexistforbirthweight,birthlength,abdominal
Horanetal.NutritionJournal2014,13:78 Page5of12
http://www.nutritionj.com/content/13/1/78
Table1Generalmaternalcharacteristicsduringpregnancy,neonatalanthropometryandcomparisonofcontroland
intervention(lowglycaemicindexdiet)groups
n Intervention Control Total p-value
MotherAge(yrs) 542 32.83±3.97 32.91±3.91 32.87±3.93 0.824
MotherHeight(cm) 542 165.58±12.29 165.03±11.54 165.27±11.86 0.591
MotherWeight(kg) 542 72.45±12.95 72.32±12.97 72.38±12.94 0.905
MotherBMI1(kg/m2) 542 26.19±4.35 26.38±4.42 26.30±4.38 0.616
Gestationalweightgain(kg) 273 12.48±4.40 14.13±4.55 13.39±4.55 0.003
Postprandialglucose* 537 6.47±1.42 6.77±1.77 6.64±1.63 0.031
Motherwell-being%score 508 56.37±15.38 60.21±15.19 58.48±15.37 0.005
Daysperweekwalking≥30min2 426 3.55±1.86 3.44±1.74 3.48±1.79 0.527
Moderateactivity(minperweek) 308 70.80±49.70 61.80±36.64 66.20±43.46 0.066
T1EnergyIntake(kcal/d) 521 1828.25±407.98 1874.02±474.10 1854.47±446.12 0.245
T2EnergyIntake(kcal/d) 529 1803.43±440.82 1943.24±476.75 1883.50±467.54 0.001
T3EnergyIntake(kcal/d) 541 1832.55±424.02 1932.42±472.56 1889.62±454.91 0.011
T1ProteinIntake(%TE) 521 17.23±3.04 16.80±3.10 16.98±3.08 0.114
T2ProteinIntake(%TE) 529 17.79±3.16 16.75±2.92 17.21±3.07 <0.001
T3ProteinIntake(%TE) 541 17.62±3.18 16.69±2.96 17.10±3.09 <0.001
T1CarbohydrateIntake(%TE) 521 50.31±6.65 50.32±6.35 50.31±6.47 0.996
T2CarbohydrateIntake(%TE) 529 48.99±5.91 49.88±5.94 49.48±5.94 0.089
T3CarbohydrateIntake(%TE) 541 49.05±5.50 49.96±6.09 49.57±5.85 0.072
T1TotalFatIntake(%TE) 521 35.47±5.92 35.83±5.45 35.68±5.66 0.471
T2TotalFatIntake(%TE) 529 36.11±5.35 36.30±5.43 36.22±5.39 0.673
T3TotalFatIntake(%TE) 541 36.19±5.26 36.20±5.40 36.19±5.33 0.978
T1SFA3Intake(%TE) 521 13.35±3.00 13.76±2.93 13.57±2.97 0.119
T2SFA3Intake(%TE) 529 13.45±3.00 13.95±3.00 13.72±3.02 0.055
T3SFA3Intake(%TE) 541 14.00±3.17 13.90±3.10 13.93±3.14 0.721
T1MUFA4Intake(%TE) 521 11.09±2.55 11.38±2.40 11.26±2.47 0.184
T2MUFA4Intake(%TE) 529 11.41±2.37 11.49±2.26 11.46±2.31 0.695
T3MUFA4Intake(%TE) 541 11.25±2.29 11.38±2.21 11.33±2.24 0.497
T1PUFA5Intake(%TE) 521 5.99±2.10 5.81±2.11 5.90±2.12 0.347
T2PUFA5Intake(%TE) 529 6.07±1.76 5.78±1.80 5.91±1.80 0.066
T3PUFA5Intake(%TE) 541 5.79±1.81 5.72±1.87 5.76±1.85 0.672
T1GlycaemicIndex 521 57.38±4.24 57.71±4.03 57.56±4.12 0.365
T2GlycaemicIndex 529 56.26±4.04 57.83±3.71 57.13±3.93 <0.001
T3GlycaemicIndex 541 56.12±3.88 57.70±3.88 57.00±3.96 <0.001
T1GlycaemicLoad 521 132.40±32.79 136.45±38.70 134.69±36.21 0.209
T2GlycaemicLoad 529 123.34±31.42 140.26±36.90 132.93±35.62 <0.001
T3GlycaemicLoad 541 126.49±30.03 139.81±37.22 134.10±34.96 <0.001
NeonatalWeight(kg) 542 4.05±0.47 4.01±0.45 4.02±0.46 0.274
NeonatalLength(cm) 542 53.01±2.40 52.48±2.68 52.73±2.56 0.160
NeonatalAbdominalCirc6(cm) 222 33.24±2.29 33.54±1.96 33.40±2.11 0.303
NeonatalWaistCirc6:HeightRatio 182 0.63±0.04 0.64±0.05 0.64±0.05 0.013
NeonatalSOSF7(mm) 186 28.66±5.60 28.52±4.88 28.58±5.18 0.848
NeonatalSumTSF8andSSF9(mm) 186 13.87±2.83 13.92±2.60 13.90±2.69 0.906
NeonatalTSF8:SSF9Ratio 186 1.01±0.19 1.00±0.18 1.00±0.19 0.489
Horanetal.NutritionJournal2014,13:78 Page6of12
http://www.nutritionj.com/content/13/1/78
Table1Generalmaternalcharacteristicsduringpregnancy,neonatalanthropometryandcomparisonofcontroland
intervention(lowglycaemicindexdiet)groups(Continued)
NeonatalWaist:HipCirc6Ratio 221 0.99±0.06 1.00±0.05 1.00±0.06 0.192
“n”denotesnumber,%TEdenotespercentageoftotalenergy,p<0.05wasconsideredstatisticallysignificant(2-tailedsignificancegeneratedfromANOVAor
t-testsasappropriate).
*1hourpost50gglucosechallengetest.
Abbreviations:1BMI,bodymassindex;2min,minutes;3SFA,saturatedfattyacids;4MUFA,monounsaturatedfattyacids;5PUFA,polyunsaturatedfattyacids;6circ,
circumference;7SOSF,sumofskinfolds;8TSF,tricepsskinfold;9SSF,subscapularskinfold.
Table2Maternalandpaternalcharacteristicsandmaternalnutrientintakesassociatedwithneonatalanthropometry–
adjustedformaternalpre-pregnancyBMI,educationlevel,lengthofgestationandneonategender1,2
B SEB p-value R2adj F p-value
BirthWeight
GestationalweightGain 23.220 6.040 0.000 0.233 11.547 0.000
Length
Motherbaselinesmoker −6.220 1.490 0.000 0.278 6.193 0.000
AbdominalCircumference
Trimester3SFA3(%TE) 0.147 0.050 0.004 0.055 2.484 0.019
Trimester3PUFA4(%TE) −0.184 0.099 0.065
ThighCircumference
Baselinestrenuousphysicalactivity −0.706 0.257 0.014 0.467 4.365 0.008
ChestCircumference
Maternalweightbooking 0.150 0.080 0.080 0.377 2.906 0.039
Baselinestrenuousphysicalactivity −0.820 0.286 0.012
Subscapskinfoldthickness
Trimester3PUFA4(%TE) −0.140 0.074 0.062 0.050 2.301 0.038
Subscapular:Tricepsskinfoldratio
well-being%score −0.003 0.001 0.006 0.051 2.410 0.030
WaistCircumference:LengthRatio
Maternalage −0.002 0.001 0.010
Baselinemothersmoker 0.083 0.018 0.000
Baselinemotherarmcircumference 0.005 0.002 0.046
Trimester2glycaemicload 0.003e−1 0.000 0.059 0.381 8.003 <0.001
Postprandialglucoseat28weeks 0.007 0.003 0.036
Trimester3SFA3(%TE) 0.004 0.001 0.001
Group 0.019 0.008 0.021
WaistCircumference:LengthRatio(adjustedfordefiniteunderreporting)
Maternalage −0.002 0.001 0.008
Baselinemothersmoker 0.096 0.018 <0.001
Baselinemotherarmcircumference 0.005 0.002 0.025 0.401 8.594 <0.001
Postprandialglucoseat28weeks 0.006 0.003 0.050
Trimester3SFA3(%TE) 0.003 0.001 0.005
Group 0.028 0.010 0.007
1Well-establishedinfluencesonneonatalsize,2Groupaffiliationwasalsoincludedinallmodelsasthisstudywasarandomisedcontroltrial.%TEdenotes
percentageoftotalenergy.Multipleregressionanalysiswasusedinthisanalysisandonlystatisticallysignificantoverallmodelswereincluded.p<0.05was
consideredstatisticallysignificant.
Abbreviations:3SFA,saturatedfattyacids;4PUFA,polyunsaturatedfattyacids.
Horanetal.NutritionJournal2014,13:78 Page7of12
http://www.nutritionj.com/content/13/1/78
Table3Maternalandpaternalcharacteristicsandmaternalnutrientintakesassociatedwithneonatal
anthropometry-unadjustedanalysis
B SE p-value R2
BirthWeight
Maternalweightbooking(kg) 6.506 1.286 0.000 0.032
Maternalheight(cm) 3.548 1.488 0.017 0.006
Paternalheight(cm) 7.458 2.838 0.009 0.012
Maternalmid-upperarmcircumference(cm) 18.656 5.223 0.000 0.016
Gestationalweightgain(kg) 15.676 4.863 0.001 0.025
BMI1booking(kg/m2) 11.752 3.717 0.002 0.012
BirthLength
Paternalheight(cm) 0.076 0.038 0.049 0.024
Attendgym(y/n) −1.193 0.531 0.026 0.021
Motherbaselinesmoker(y/n) 2.724 0.964 0.005 0.031
Motherlivingwithpartner(y/n) −1.350 0.507 0.008 0.027
Trimester2protein(%TE) 0.131 0.055 0.017 0.025
Trimester2carbohydrate(%TE) −0.068 0.032 0.034 0.019
AbdominalCircumference
Maternalweightbooking(kg) 0.021 0.010 0.030 0.014
Trimester1SFA2(%TE) 0.100 0.047 0.035 0.015
Trimester3SFA2(%TE) 0.118 0.042 0.006 0.030
Trimester3PUFA3(%TE) −0.215 0.080 0.008 0.027
ThighCircumference
Group(interventionvscontrol) 0.404 0.200 0.044 0.012
Baselinestrenuousphysicalactivity(freq/wk) −0.851 0.295 0.008 0.220
ChestCircumference
Maternalweightbooking(kg) 0.025 0.012 0.036 0.013
Baselinestrenuousphysicalactivity(freq/wk) −0.882 0.356 0.020 0.166
Baselineattendgym(y/n) −1.170 0.505 0.021 0.019
HipCircumference
Maternalweightbooking(kg) 0.030 0.010 0.004 0.028
Maternalmid-upperarmcirc4booking(cm) 0.125 0.043 0.004 0.029
MaternalBMI1booking(kg/m2) 0.082 0.030 0.007 0.024
Mid-UpperArmCircumference
Baselinestrenuousphysicalactivity(freq/wk) −0.511 0.226 0.033 0.137
Trimester2protein(%TE) −0.060 0.030 0.046 0.013
Subscapularskinfoldthickness
Trimester3PUFA3(%TE) −0.163 0.062 0.009 0.030
Tricepsskinfoldthickness
Trimester1GlycaemicIndex −0.067 0.032 0.037 0.018
Trimester3PUFA3(%TE) −0.142 0.064 0.027 0.021
Bicepsskinfoldthickness
Livingwithpartner(y/n) 0.569 0.284 0.046 0.014
Trimester3PUFA3(%TE) −0.125 0.061 0.042 0.017
SumofAllskinfoldthicknesses
Trimester3PUFA3(%TE) −0.512 0.217 0.019 0.024
Horanetal.NutritionJournal2014,13:78 Page8of12
http://www.nutritionj.com/content/13/1/78
Table3Maternalandpaternalcharacteristicsandmaternalnutrientintakesassociatedwithneonatal
anthropometry-unadjustedanalysis(Continued)
Waist:HipCircumferenceRatio
Baselinemildphysicalactivity(freq/wk) −0.004 0.002 0.038 0.018
Timester2SFA2(%TE) 0.003 0.001 0.028 0.017
SumofTricepsandSubscapularskinfoldthickness
Trimester3PUFA3(%TE) −0.305 0.112 0.007 0.033
Subscapular:Tricepsskinfoldthicknessratio
Percentagemoodscore −0.003 0.001 0.010 0.029
WaistCircumference:Lengthratio
Maternalage(yrs) −0.002 0.001 0.023 0.022
Baselinestrenuousphysicalactivity(freq/wk) −0.019 0.007 0.013 0.232
Baselinesmoker(y/n) −0.072 0.017 0.000 0.070
Maternalmid-upperarmcirc4booking(cm) 0.002 0.001 0.031 0.018
Trimester2glycaemicload 0.000 0.000 0.005 0.039
Postprandialglucoseat28weeks(mmol/l) 0.004 0.002 0.018 0.021
Trimester2%energyfromprotein(%TE) −0.002 0.001 0.016 0.026
Trimester2%energyfromSFA2(%TE) 0.002 0.001 0.050 0.016
Trimester3SFA2(%TE) 0.003 0.001 0.006 0.035
Trimester3PUFA2(%TE) −0.005 0.002 0.020 0.024
“n”denotesnumber,%TEdenotespercentageoftotalenergy.SimpleLinearRegressionwasusedinthisanalysisandonlystatisticallysignificantassociationswere
includedinthistable.p<0.05wasconsideredstatisticallysignificant.Therewerenostatisticallysignificantassociationswithneonatalthighskinfoldthickness,
thereforethisvariableisnotincludedintheabovetable.
Abbreviations:1BMI,bodymassindex;2SFA,saturatedfattyacids;3PUFA,polyunsaturatedfattyacids;4circ,circumference.
circumference, thigh circumference, chest circumference, significantly negatively associated with maternal age
subscapularskinfoldthickness,subscapular:tricepsskin- andpositivelyassociatedwithmaternalsmoking,mater-
foldthicknessratioandwaistcircumference:lengthratio nal mid-upper arm circumference (MUAC) in early
asfollows: pregnancy,SFAintakeintrimester3,postprandialglucose
Birthweight was positively associated with gestational at 28 weeks gestation and membership of the control
weight gain R2 23.3% (F=11.547, p<0.001). Birthlength group and showed a trend towards a positive association
adj
wasnegativelyassociatedwithmaternalsmokingR2 27.8% with Glycaemic Load in trimester 2 R2 38.1% (F=8.003,
adj adj
(F=6.193, p<0.001). Neonatal abdominal circumference p<0.001).
was positively associated with maternal saturated fatty
acid(SFA)intakeandshowedanegativetrendtowardsas- Macronutrientintakes
sociationwithpolyunsaturatedfattyacid(PUFA)intakein Maternal macronutrient intakes are displayed in Table 1.
trimester 3 R2 5.5% (F=2.484, p=0.019). Neonatal thigh Their association with neonatal anthropometric mea-
adj
circumferencewasnegativelyassociatedwithfrequencyof surements using simple linear regression are displayed
strenuous physical activity reported in early pregnancy in Table 3 and those that remained significantly associ-
R2 46.7%(F=4.365,p=0.008).Neonatalchestcircumfer- ated with neonatal anthropometry when analysed using
adj
ence showed a trend towards a positive association with multiple linear regression are displayed in Table 2. SFA
maternal weight at booking and was significantly nega- intake in trimester 3 was positively associated with neo-
tively associated with frequency of strenuous physical natal abdominal circumference (B=0.147, p=0.004)
activity reported in early pregnancy R2 37.7% (F=2.906, while there was a trend towards PUFA intake in trimes-
adj
p=0.039). Neonatal subscapular skinfold thickness was ter 3 being negatively associated (B=−0.184, p=0.065).
showedatrendtowardsanegativeassociationwithPUFA There was a trend towards PUFA intake in trimester 3
intake in trimester 3 R2 5.0% (F=2.301, p=0.038). being negatively associated with subscapular skinfold
adj
Neonatal subscapular:triceps skinfold thickness ratio, a thickness (B=−0.140, p=0.062). Finally, SFA intake in
measureofcentraladiposity,wassignificantlynegatively trimester3waspositively associated with abdominal adi-
associated with percentage well-being score R2 5.1% posity as measured by waist circumference:length ratio
adj
(F=2.410, p=0.030). Neonatal waist circumference: while there was a trend towards a positive association
length ratio, another measure of central adiposity, was withGLintrimester2(B=0.004,p=0.001andB=0.003e−1,
Horanetal.NutritionJournal2014,13:78 Page9of12
http://www.nutritionj.com/content/13/1/78
p=0.059 respectively). There was no difference in maternal ponderal index in the literature [55,56], we were unable
macronutrient intake between the control and intervention to identify any studies that had measured waist:length
groups exceptfor maternal GIand GL intake and protein ratio atbirth in relation to maternal smoking status. The
intakeas a percentage of total energy intake as previously only studies identified that had measured waist:length
described [27]. No significant association was found be- ratio at birth were from the same group and involved
tween level of concordance with a DASH diet and any of creation of normative waist:length centile charts at birth
the neonatal body measurements examined. Similarly, [57]. The increased waist:length ratio observed in the
previously defined healthy and unhealthy diet clusters off-spring of smokers in this study appears to reflect the
werenotassociatedwithneonatalbodymeasurements. reduced height also observed and likely indicates that,
while birthweight is not increased, central and visceral
Discussion adiposity may be, increasing the risk of metabolic syn-
Themainmaternalfactorassociatedwithincreasedbirth drome inlaterlife.
weight was greater gestational weight gain while the Maternal early pregnancy MUAC was positively asso-
mainmaternalfactorassociatedwithgreater birthlength ciated with neonatal waist:length ratio. MUAC has been
was non-smoking status. Neonatal central adiposity, de- found to be well correlated with maternal weight and
termined using waist:length ratio, was negatively associ- BMI and remains stable in pregnancy i.e. unaffected by
ated with maternal age, and positively associated with length of gestation [58]. As a measure of maternal over-
maternal smoking status, pre-pregnancy mid-upper arm weight and obesity, MUAC has been found to be posi-
circumference,trimester3saturatedfatintake,postpran- tively associated with birthweight [59-61]. Again, thereis
dial glucose at 28 weeks gestation and membership of no information on its association with neonatal waist:
the control group and showed a trend towards a positive length ratiotodate tothebest ofourknowledge.
associationwithtrimester2glycaemic load. Trimester 3 SFA intake was positively associated with
Similar to other studies [45-47], gestational weight abdominal circumference and waist:length ratio. Trimes-
gain was found to be positively associated with birth- ter 3 PUFA intake showed a trend towards a negative as-
weight in this cohort. Guidelines from the Institute of sociation with abdominal circumference and subscapular
Medicine (IOM) rely on pre-pregnancy BMI to deter- skinfold thickness. There is a paucity of similar data into
mine an appropriate range of gestational weight gain the effect ofmaternalfat intake in pregnancy onneonatal
[48]. It is well established that those who exceed the adiposity.Highfatisocaloricdietinratshasbeenfoundto
IOM guidelines are at risk of delivering a macrosomic result in no difference in birthweight of pups [62]. Simi-
infant [49]. Excess gestational weight gain increases the larly,Brionetal.[63]foundnoassociationbetweenmater-
normal insulin resistance that occurs in pregnancy and nal diet and child adiposity at 9 or 11 years of age in
may also affect other hormones that regulate nutrient humans.However,thequalityofdietarymacronutrientin-
transport across the placenta resulting in increased fetal takes may be more important than absolute intakes and
insulin secretion, growth and adiposity [45]. Research by high SFA diet in pregnancy has been found by Murrin
Ludwig et al. into multiparous women throughout suc- et al. [64] to be positively associated with child weight at
cessive pregnancies has found that gestational weight age5.Maternalintakeoftransfattyacidsintrimester2of
gainisresponsibleforincreasedbirthweightdespite con- pregnancy was also found by one group to be positively
trolling forgeneticandsociodemographic factors [45]. associated with birthweight [65] while other studies have
Our finding that maternal smoking during pregnancy reportedconflictingresultspossiblyduetoexaminationof
was associated with decreased birth length is well estab- different trans fatty acids [17,66,67]. Fetal fat deposition
lished in the literature [50,51]. Maternal smoking was increases with gestational age, therefore our observation
also found to be positively associated with waist:length that maternal fatty acid intake in trimester 3 is associated
ratio, the equivalent of which (waist:height ratio) has with neonatal adiposity is reasonable [3]. Although there
been found to be a good measure of central adiposity in is little research into the effect of SFA in pregnancy there
adults and children with a ratio of ≥0.5 indicating excess is more interest into the effect of maternal dietary PUFA
central adiposity [52]. A recent study by Brambilla et al. intake, in particular the possible anti-obesogenic effect of
found it to be a better measure of adiposity than waist agreateromega3:omega6ratio.However,arecentreview
circumference or BMI in children and adolescents [53]. by Hauner et al. [68] found that there is little evidence of
A meta-analysis by Ino [54] found that maternal smok- this to date due to conflicting study results. Results of a
ing is associated with childhood overweight and obesity, prospective intervention study involving omega 3 supple-
possibly through a combination of the thrifty phenotype mentation and dietary reduction of omega 6 showed no
and catch-up growth during early infancy. Although ma- effect on fat mass at age 1 [69]. This study, the INFAT
ternal smoking during pregnancy is associated with re- study, similar to others involving fish oil supplementation
duced birth weight, length and relatively unchanged found that birthweight and length of gestation were
Horanetal.NutritionJournal2014,13:78 Page10of12
http://www.nutritionj.com/content/13/1/78
increased, however adiposity was not affected [69]. We for the full cohort, however weight and length measure-
believe ours is the first study to show an association ments were taken for all infants. It should also be noted
betweenneonatalcentraladiposityanddietaryfatquality. that this was a cohort at risk of macrosomia, therefore
Trimester 2 Glycaemic load (GL) was included in the care should be taken regarding generalisation of results
multiple linear regression model and showed a trend to- to other populations. A further limitation was that fatty
wards a positive association with waist:length ratio. How- acid composition was not broken down into omega 3,
ever, while the overall model, which included trimester 2 omega 6 and trans fats. This clinical trial was originally
GL, was statistically significantly associated with waist: focused on the effect of GI, GL and glycaemic control
lengthratio,T2GLwasnotindependentlysignificantlyas- on birthweight but nevertheless, data on SFA, MUFA,
sociated and lost significance when underreporting was PUFA and total fat was available and this is the first hu-
controlled for. Maternal postprandial blood glucose and man study to report an association between dietary fat
membership of the control group were significantly posi- qualityandneonatalcentral adiposity.
tively associated with waist:length ratio although the rela-
tionship between postprandial glucose and waist:length Conclusion
circumference was reduced to a trend (p= 0.050) when Several maternal diet and lifestyle factors were associated
underreporting was controlled for. There has been much with neonatal body composition. The finding that neo-
researchintoglycaemiccontrolandintotheeffectofdiet- natal central adiposity was positively associated with
ary glycaemic index and load during pregnancy on birth- maternal dietary saturated fat and showed a negative
weight due to the well-established risk of macrosomia in trend with polyunsaturated fat and a significant negative
gestational diabetes [70]. The risk of macrosomia and in- association with membership of the low GI intervention
creased neonatal adiposity has also been found to be in- grouphighlightstheimportanceofdietaryqualityinpreg-
creased towards the upper limits of normalblood glucose nancy and the need for further research and education in
controlinpregnancy[9,71,72]whichisinlinewithourre- thisarea.
sults. Low glycaemic index diet has been found to ameli-
Competinginterests
orate the normal pregnancy-related increase in glucose
Theauthorsdeclarethattheyhavenocompetinginterests.
intolerance associated with pregnancy resulting in fewer
peaks in maternal postprandial glucose concentration [9]. Authors’contributions
Thepositivetrend towardsassociationbetweenpostpran- MKH-carriedoutdataanalysis,assignedtheDASHdietscoreandwrotethe
manuscript,CAMcG-carriedoutdataentryandk-meansclusteranalysis,ERG
dial glucose and waist:length ratio in this study may indi-
gaveadviceonstudydesignandeditingofthemanuscript,JMDcarriedout
cate that the reduction in postprandial blood glucose manuscriptediting,FMMcAwasresponsibleforconceptionofthestudy,advice
levels observed with low glycaemic index diet was associ- onstudydesignandmanuscriptediting.Allauthorsreadandapprovedthe
finalmanuscript.
ated with reduced glucose transfer to the fetus and there-
fore less deposition of fetal adiposity. While the ROLO Acknowledgements
study found that the intervention group reduced the GI ThisstudywassupportedbytheHealthResearchBoardIreland,theHealth
ResearchCentreforHealthandDietResearch,TheNationalMaternityHospital
and GL of their diet, even when underreporters were ex- MedicalFundandtheEuropeanUnion’sSeventhFrameworkProgramme
cludedfromtheanalysis[27],andhadreducedglucosein- (FP7/2007-2013),projectEarlyNutritionundergrantagreementno.289346.
tolerance, no difference in birthweight was observed. In TheauthorswouldliketothankJacintaByrne,researchmidwife,whocarried
outdatacollectionandRicardoSeguradowhogavestatisticaladvice.
contrast,arecentretrospectiveanalysisofmembersofthe
Danish NationalBirth Cohort has found that those in the Authordetails
highest GL quintile had significantly higher offspring 1UCDObstetricsandGynaecology,SchoolofMedicineandMedicalScience,
UniversityCollegeDublin,NationalMaternityHospital,Dublin2,Ireland.
birthweight than those in the lowest quintile [73] indicat- 2ScienceCentre–South,UniversityCollegeDublinSchoolOfAgriculture&
ing that effects may be observed at extremes. The finding FoodScience,Belfield,Dublin4,Ireland.
by the current study, that neonatal waist:length ratio was
Received:3April2014Accepted:8July2014
lower in the intervention group indicates that improved
Published:1August2014
dietary carbohydrate quality may be associated with re-
ducedcentral adiposityratherthanbirthweightatlessex- References
1. BussC,EntringerS,SwansonJM,WadhwaPD:TheRoleofStressinBrain
treme levels as the reduction in GI and GL observed in Development:TheGestationalEnvironment’sLong-TermEffectsonthe
thisstudywasquitemodest. Brain.InCerebrum:theDanaforumonbrainscience.NewYork,NY:Dana
This was a large, well-powered clinical trial designed Foundation;2012.
2. RitzE,AmannK,KoleganovaN,BenzK:Prenatalprogramming—effectson
to examine the effect of diet and lifestyle on neonatal bloodpressureandrenalfunction.NatRevNephrol2011,7:137–144.
sizeandadiposity.Richdietarydatawasavailableateach 3. HaggartyP:Fattyacidsupplytothehumanfetus.AnnuRevNutr2010,
trimester in combination with biochemical measures of 30:237–255.
4. HeerwagenMJ,MillerMR,BarbourLA,FriedmanJE:Maternalobesityand
glycaemic control. One limitation of this study was that
fetalmetabolicprogramming:afertileepigeneticsoil.AmJPhysiolRegul
detailed neonatal anthropometric data was not available IntegrCompPhysiol2010,299:R711–R722.