Maternal Adipose Tissue Expansion, A Missing Link in the Prediction of Birth Weight Centile

Abstract

Context

Maternal body mass index (BMI) is associated with increased birth weight but does not explain all the variance in fetal adiposity.

Objective

To assess the contribution of maternal body fat distribution to offspring birth weight and adiposity.

Design

Longitudinal study throughout gestation and at delivery.

Setting

Women recruited at 12 weeks of gestation and followed up at 26 and 36 weeks. Cord blood was collected at delivery.

Patients

Pregnant women (n = 45) with BMI 18.0 to 46.3 kg/m² and healthy pregnancy outcome.

Methods

Maternal first trimester abdominal subcutaneous and visceral adipose tissue thickness (SAT and VAT) was assessed by ultrasound.

Main Outcome Measures

Maternal body fat distribution, maternal and cord plasma glucose and lipid concentrations, placental weight, birth weight, and fetal adiposity assessed by cord blood leptin.

Results

VAT was the only anthropometric measure independently associated with birth weight centile (r² adjusted 15.8%, P = .002). BMI was associated with trimester 2 and trimesters 1 through 3 area under the curve (AUC) glucose and insulin resistance (Homeostatic Model Assessment). SAT alone predicted trimester 2 lipoprotein lipase (LPL) mass (a marker of adipocyte insulin sensitivity) (11.3%, P = .017). VAT was associated with fetal triglyceride (9.3%, P = .047). Placental weight was the only independent predictor of fetal adiposity (48%, P < .001). Maternal trimester 2 and AUC LPL were inversely associated with fetal adiposity (r = -0.69, P = .001 and r = -0.58, P = .006, respectively).

Conclusions

Maternal VAT provides additional information to BMI for prediction of birth weight. VAT may be a marker of reduced SAT expansion and increased availability of maternal fatty acids for placental transport.

Pregnancy, Adipose tissue, Birth weight, Insulin resistance, Body fat distribution

Introduction

Maternal obesity occurs in around 20% of the United Kingdom antenatal population (1) and is associated with an increased risk of adverse pregnancy outcome (2) including the metabolic diseases of pregnancy, gestational diabetes mellitus (GDM), and preeclampsia. Maternal obesity is also associated with offspring obesity, both at birth and in later life (3). Studies show that maternal prepregnancy total body fat predicts birth weight, though it is unclear to what extent this is explained by fetal somatic growth or fetal accumulation of fat (4). There is increasing concern about the effect of maternal body mass index (BMI) on the long-term metabolic health of the fetus (5).

Maternal obesity is associated with maternal insulin resistance (6) and metabolic dysfunction (7,8). GDM is defined by maternal hyperglycemia, and the associated fetal macrosomia may be explained by increased placental transport of glucose leading to increased fetal insulin secretion and hence increased growth, as described by Pedersen (9). This observation has now been extended to glucose-tolerant mothers, as in the Hyperglycaemia and Adverse Pregnancy Outcome trial, which highlighted the importance of maternal plasma glucose concentrations for increased birth weight and adiposity in the offspring (5). However, plasma glucose concentration did not explain all the variance in offspring adiposity, with residual contributions coming from maternal prepregnancy BMI and gestational weight gain (10–12). Furthermore, increased birth weight is observed even in GDM pregnancies in which plasma glucose is well-controlled (10,13,14).

There is evidence to suggest that high maternal fat mass and insulin resistance may expose the fetus to fuels other than glucose that could contribute to higher birth weight (13,15-17). Maternal hypertriglyceridemia is a key element of maternal obesity and insulin resistance (7,8). Whole-body lipolysis is increased in the third trimester of pregnancy (18) and is associated with maternal fat mass and estimated fetal weight (19). Maternal plasma triglyceride and free fatty acid also correlate with birth weight and measures of neonatal adiposity in GDM and in women screened for GDM with normal glucose tolerance (20–22). However, it is not clear if these relationships exist in healthy nonobese pregnancy, to what extent they are independent of maternal body weight, and whether measures of maternal body fat distribution may be superior predictors of fetal birth weight and/or adiposity (23).

Obesity, as defined by BMI in excess of 30, is a universally accepted measure of body fatness, but BMI conveys no information about the quantity, quality, location, or metabolic function of discrete fat depots. BMI is a relatively weak proxy for discriminating metabolic dysfunction and cardiometabolic risk in comparison to central obesity, especially when the latter is distinguished by high intra-abdominal visceral adipose tissue (VAT) (24). BMI is unable to distinguish between individuals (pregnant or nonpregnant) who store fat as relatively benign subcutaneous adipose tissue (SAT) or as VAT, which is intimately associated with insulin resistance, hyperglycemia, hypertriglyceridemia, metabolic dysfunction and pathology (25–27). None of these studies that related maternal adiposity to offspring birthweight and adiposity assessed body fat distribution.

In studies of pregnancy that did assess body fat distribution, preperitoneal and visceral fat were shown to increase during gestation, whereas SAT declined (28–32). Measures of VAT strongly predict metabolic complications of pregnancy, GDM (33,34) and preeclampsia (35,36), but it is not clear to what extent VAT is directly related to insulin resistance and increased plasma lipids and glucose in pregnancy (33,37-42). Although VAT, measured by ultrasound between 12 and 20 weeks of gestation, is associated with fetal growth in overweight and obese women in the first trimester (43), with fetal growth and adiposity in the second trimester (44), and birthweight (45), the effect of abdominal SAT thickness on these measures has been largely overlooked. SAT thickness is correlated with first trimester fetal growth, but less so than VAT (43). SAT has been found to be predictive (46–48) and nonpredictive of GDM (49), and is also less strongly associated with metabolic risk factors in pregnancy than VAT (37).

The aim of this study was to assess the contribution of maternal body fat distribution to offspring birth weight and maternal and fetal insulin resistance to advance our understanding of the consequences of maternal obesity. Maternal first trimester measures of adiposity including BMI, VAT, abdominal SAT, and hip circumference (a biomarker of lower body SAT) were assessed as predictors of birth weight, adiposity, and metabolic dysfunction in neonates of healthy pregnant women. We also determined if any relationship could be explained by the influence of these fat depots on maternal glucose and lipid metabolism or placental weight.

Methods

Longitudinal study of pregnancy

Sixty women registered for obstetric care at the Princess Royal Maternity Unit, Glasgow, who were healthy and normotensive with no significant medical history were recruited and followed prospectively throughout pregnancy. This study was initially designed to assess the impact of maternal obesity on microvascular function and powered for that outcome (50). Seven women were excluded from the final analysis; 3 delivered preterm, 2 were excluded because of missing birth weight data, 1 had a miscarriage, and 1 had a baby with DiGeorge syndrome. All remaining women had no pregnancy-related complications. Baseline data and complete longitudinal data were available for 53 and 45 women, respectively. The study was performed according to the Declaration of Helsinki, approval was granted by the Research Ethics Committee of North Glasgow University NHS Trust, and each subject gave written informed consent. The women attended after an overnight fast (>10 hours). Blood samples were collected at a mean of 12.4 (range, 8-14) trimester 1, 26.1 (range, 24–28) trimester 2, and 35.5 (range, 33–38) trimester 3 weeks of gestation. Patient characteristics were recorded at the first antenatal hospital appointment and delivery details from patient notes. Customized birth weight centiles were calculated using the Gestation Network Centile Calculator 5.4 (http://www.gestation.net/birthweight_centiles/centile_online.htm). Deprivation category score, a measure of socioeconomic status, was assigned using the Scottish Area Deprivation Index for Scottish postcode sectors, 1998 (51). Placental tissue and fetal cord blood was collected at delivery from a subgroup of these pregnancies (n = 23, 42%).

Baseline anthropometric and fat thickness measurements

At the first antenatal appointment (mean, 12.4 weeks of gestation), patient height, weight, waist circumference, and hip circumference were measured. Waist circumference was measured at the level of the umbilicus. Hip circumference was measured at the widest point over the buttocks. Waist and hip circumference were measured in duplicate to the nearest 0.5 cm. If the difference between the 2 measurements was greater than 2 cm, a third measurement was taken and the mean of the 2 closest measurements was calculated. All measurements were taken by the same examiner. BMI was calculated as booking weight (kg), divided by height (m) squared. Baseline upper body abdominal SAT and upper body VAT thickness was assessed by ultrasound (52). Measurements were taken 2 cm below the xiphisternum, and the abdominal probe was placed on the skin with minimal pressure. Abdominal SAT was measured as the thickness between the inferior border of the dermal layer and the rectus abdominus sheath at the level of the umbilicus. Visceral fat was taken as the vertical measurement between the rectus sheath and the aorta at the umbilicus. Three consecutive measurements in millimeters were taken and an average reading was calculated. All measures were made by the same operator (F.S.) on the same machine.

Blood parameters

Glucose assays (53) were performed by Clinical Biochemistry, Glasgow Royal Infirmary, and plasma total cholesterol, high-density lipoprotein (HDL) cholesterol and total triglyceride concentrations were determined as described previously (7). Insulin (Mercodia) and leptin (R & D Systems) analyses were performed by ELISA according to the manufacturer’s instructions. Lipoprotein lipase (LPL) mass was determined by ELISA using bovine LPL as standard (54,55). Homeostatic Model Assessment (HOMA) was calculated as (fasting insulin [mU/L] × fasting glucose [mmol/L])/22.5). Erythrocyte membrane phospholipid fatty acid composition was measured as previously described (56), and the ratio of 16:0/18:2 n-6 was used as an index of de novo lipogenesis (57,58).

Statistical analysis

Continuous variables are represented as means (SD), categorical variables as number and percentage. Total area under the time (trimester 1 to trimester 3 weeks’ gestation) × concentration curve (AUC) were calculated using the trapezium method (59). Normality testing was carried out using the Ryan-Joiner test and data were log or square root transformed to achieve a normal distribution as necessary. Associations between variables were examined by Pearson’s correlation analysis. Two measures of gestational fuel exposure were used. First, the total area under the trimester 1 to trimester 3 maternal glucose, and triglyceride concentration and HOMA curves were calculated to assess total gestational exposure. Second, because the univariate associations between maternal anthropometrics and maternal plasma triglyceride, glucose and HOMA were the strongest and notably distinct in trimester 2 (Supplemental Fig. 1 located in a digital research material repository (60)), trimester 2 data were selected for further study. Stepwise regression analysis, a method of fitting regression models in which the choice of predictive variables and simultaneous removal of unimportant variables is carried out by an automatic procedure, was used to test associations between maternal anthropometrics and maternal or fetal blood glucose and lipids and the influence of confounding variables such as placental weight, using P-to-enter and P-to-stay P < .15. All statistical analysis was carried out in Minitab Vs18.

Results

Maternal characteristics

Demographic data for all women with first trimester measurements are shown in Table 1. Women were on average 28 years of age, had a mean BMI of 28 kg/m²; just over one-third were currently smoking during their pregnancy and just less than one-half were in their first pregnancy. More than one-half the women were classed as having deprived social status. The women had normal antenatal appointment blood pressure, had no pregnancy-related complications, and all delivered healthy babies at term. The demographic characteristics of the subgroup, where repeated longitudinal measures were available, were similar to the total group (Table 1).

Table 1.

Maternal Antenatal Booking Characteristics

	Baseline Data n = 53	Longitudinal Data n = 45	Cord Blood Data n = 23
Booking visit
Age (y)	28.3 (5.1)	28.6 (5.0)	28.9 (4.6)
Smokers, number (%)	19 (35.8)	16 (35.6)	8 (34.8)
Deprivation Index
Deprivation Category Score, number (%)
Affluent (1–2)	5 (9.4)	5 (11.1)	4 (17.4)
Intermediate (3–5)	18 (34.0)	16 (35.6)	8 (34.8)
Deprived (6–7)	30 (56.6)	24 (53.3)	11 (47.8)
Primiparous, number (%)	25 (47.2)	21 (46.7)	12 (52.2)
Systolic blood pressure (mm Hg)	118 (13)	119 (11)	118 (12)
Diastolic blood pressure (mm Hg)	68 (9)	69 (9)	68 (9)
BMI (kg/m2)	28.4 (6.0)	28.1 (5.7)	26.6 (5.7)
BMI range (kg/m2)	18.0 – 46.3	18.0 – 46.3	19.0 – 46.3
Waist circumference (cm)	89.5 (14.8)	89.5 (14.8)	84.3 (11.4)
Hip circumference (cm)a	107 (15)	108 (15)	104 (15)
Waist/hip ratio	0.83 (0.08)	0.84 (0.08)	0.81 (0.07)
Visceral fat thickness (mm)	13.4 (5.6)	13.2 (5.5)	12.1 (5.8)
Subcutaneous fat thickness (mm)	25.1 (11.4)	24.5 (10.9)	22.9 (11.0)
Visceral/subcutaneous fat thickness	0.58 (0.23)	0.58 (0.24)	0.58 (0.27)
Visceral plus subcutaneous fat thickness (mm)	38.4 (15.7)	37.7 (15.1)	35.0 (15.2)
At delivery
Gestation at delivery (d)	279 (9)	281 (7)	278 (8)
Fetal sex, number (%) male	29 (55)	24 (53)	15 (65)
Placental weight (g)	763 (175)	785 (175)	788 (193)
Birth weight (g)	3606 (555)	3680 (560)	3675 (628)
Birth weight centile
Mode of delivery, number (%)	61 (31)	62 (31)	58.8 (34)
Assisted	10 (18.9)	9 (20.0)	4 (17.4)
Elective cesarean section	4 (7.5)	3 (6.7)	2 (8.7)
Emergency cesarean section	7 (13.2)	6 (13.3)	3 (13.0)
Vaginal	32 (60.4)	27 (60.0)	14 (60.9)

Values are mean and SD for continuous variables or number (%) for categorical variables.

^an = 1 missing data.

Relationship between maternal anthropometrics and offspring birth weight centile and placental weight

Maternal BMI is a recognized predictor of offspring birthweight. To test whether other, more specific, measures of maternal adiposity might be better predictors of offspring birth weight, univariate correlation between maternal anthropometric measures and birth weight centile, birth weight and placental weight were first assessed. Birth weight centile was associated with maternal BMI (r = 0.41, P = .002), waist circumference (r = 0.42, P = .003), hip circumference (r = 0.32, P = .021), SAT thickness (r = 0.34, P = .012), VAT thickness (r = 0.41, P = .003), and SAT plus VAT thickness (r = 0.39, P = .004), but not waist-hip ratio (r = 0.16, P = .25) or VAT/SAT ratio (r = 0.01, P = .97). No maternal anthropometric measures were associated with birth weight alone, and only maternal BMI was weakly associated with placental weight (r = 0.27, P = .047).

To examine possible multivariable associations between maternal anthropometrics and birth weight centile, the former was entered into a stepwise regression model (P to enter and P to stay, .15). VAT was the only anthropometric measure significantly associated with birth weight centile (r² adjusted 15.8%, P = .002, Fig. 1). A 1-mm increase in VAT thickness resulted in a 2.26 centile increase in birthweight centile. A VAT thickness of up to 10 mm was associated with the 43rd (unadjusted) and 38th (after adjustment for BMI, waist circumference, hip circumference, SAT thickness, and SAT plus VAT thickness) birth weight centile. Above 10 mm VAT, birth weight centiles were between the 70th and 80th centile. Although inclusion of placental weight in the regression model attenuated the relationship between VAT and birth weight centile, it remained significant (r² adjusted 11.8%, P = .006; birth weight centile = -12.3 + 1.891 VAT [mm] + 0.0630 placental weight [g]) suggesting an independent association between VAT thickness and birth weight centile. Placental weight was significantly associated with birth weight centile in this model (r² adjusted 12.1%, P = .005), suggesting that placental weight also has an independent contribution to birth weight equivalent to that of VAT thickness.

Figure 1.

Offspring birth weight centile by VAT thickness: 0-10 mm (n = 19, mean [SD] 7.5 [2.0] mm), 10-20 mm (n = 27, 15.0 [2.6] mm), 20-30 mm (n = 7, 22.9 [2.9] mm). Means and 95% confidence interval for the mean are plotted for unadjusted data and data adjusted for other anthropometric measures in a stepwise regression (body mass index, waist circumference, hip circumference, SAT thickness, SAT plus VAT thickness). Abbreviations: SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue.

Relationships between maternal anthropometrics and fetal cord plasma glucose and lipids

In a univariate analysis maternal VAT thickness was significantly correlated (P < .05) with fetal cord plasma total cholesterol, triglyceride, nonesterified fatty acids, and negatively with a marker of insulin resistance (HOMA), but there was no relationship with fetal cord plasma HDL cholesterol, glucose, or insulin (Table 2). Maternal SAT thickness correlated significantly with cord plasma total cholesterol. There was no relationship between maternal BMI or hip circumference and cord plasma glucose or lipids. On multivariate regression analysis, VAT showed significant independent associations with cord plasma triglyceride, nonesterified fatty acids, and cholesterol, whereas BMI showed significant negative independent associations with cord plasma insulin and HOMA (Table 2). After inclusion of mode of delivery in the model, the associations between maternal BMI and cord plasma insulin and HOMA and between VAT and cord plasma triglyceride persisted and now hip circumference was positively associated with cord plasma cholesterol (Table 2). The inclusion of placental weight in the regression model had no effect on the associations between BMI and cord plasma HOMA, VAT thickness, and cord plasma triglyceride or between hip circumference and cord plasma cholesterol maternal BMI and HOMA. The inclusion of placental weight strengthened the negative association between BMI and cord blood insulin (r² adjusted 31%, P = .009).

Table 2.

Maternal BMI, VAT Thickness, SAT Thickness, and Hip Circumference Association With Fetal Cord Plasma Markers of Glucose and Lipid Metabolism

Response	Cord Plasma Mean (SD)	Obesity Measure	Univariate Correlation	Coefficient	Contribution to Variance, Multivariable	P	Contribution to Variance, Multivariable Including MOD	P
Glucose (mmol/L)	4.30 (1.07)	BMI	-0.17	-0.0655	VAT 6.9%	.013	No modelb	-
		SAT	-0.07
		VAT	-0.34
		Hip	0.08
Insulin (mU/L)a	6.56 (6.64)	BMI	-0.36	-0.0332	BMI 15.4%	.040	BMI 15.4%	.040
		SAT	-0.37
		VAT	-0.37
		Hip	-0.36
HOMA-IRa	1.29 (1.21)	BMI	-0.38	-0.0596	BMI 20.5%	.023	BMI 20.5%	.023
		SAT	-0.35
		VAT	-0.45c
		Hip	-0.28
Triglyceride (mmol/L)a	0.72 (0.65)	BMI	0.16	0.01994	VAT 14.8%	.043	VAT 9.3%	.047
		SAT	0.30
		VAT	0.42c
		Hip	0.14
NEFA (mmol/L)a	0.19 (0.12)	BMI	0.26	0.0236	VAT 17.5%	.034	VAT 12.8%	.075
		SAT	0.26
		VAT	0.47c
		Hip	0.28
Total cholesterol (mmol/L)a	1.79 (0.65)	BMI	0.33	0.00938	VAT 14.0%	.049	Hip 16.8%	.002
		SAT	0.42c
		VAT	0.43c
		Hip	0.33
HDL cholesterol (mmol/L)	0.74 (0.18)	BMI	0.12		No modelb		-	-
		SAT	0.02
		VAT	-0.11
		Hip	0.15

Mean (SD) cord plasma levels (n = 23) of each parameter are shown below along with their univariate association (Pearson correlation) with maternal BMI, VAT, and SAT thickness and hip circumference. The relationship between maternal BMI, VAT, and SAT thickness and fetal cord plasma metabolic measures was determined by entering BMI, SAT, VAT, and hip circumference in a stepwise regression model P to enter and P to stay .15. MOD (assisted, elective cesarean section, emergency cesarean section, and vaginal delivery) was added to the models as a confounding variable. r² adjusted and P values are stated.

^aAnalysis carried out on log-transformed data.

^bNo terms were at P < .15 to be entered into the model.

^c  P < .05.

Abbreviations: BMI, body mass index; HDL, high-density lipoprotein; MOD, mode of delivery; NEFA, nonesterified fatty acid; SAT, subcutaneous adipose tissue; VAT, visceral adipose tissue.

Relationship between maternal anthropometrics and maternal plasma glucose and lipids

Maternal BMI was correlated significantly (P < .05) with maternal trimester 2 and AUC glucose and trimester 2 and AUC HOMA (Table 3). SAT thickness was correlated significantly with maternal trimester 2 HOMA only. VAT thickness was correlated with maternal trimester 2 glucose and HOMA. Hip circumference was correlated with maternal AUC glucose and trimester 2 and AUC HOMA. On stepwise regression, only maternal BMI remained significantly associated with both maternal trimester 2 and AUC glucose and HOMA (Table 3).

Table 3.

Maternal BMI, VAT and SAT, and Hip Circumference Associations With Maternal Markers of Gestational Fuel Exposure

Response	Maternal Plasma Mean (SD)	Obesity Measure	Univariate Correlation	Coefficient	Contribution to Variance, Multivariable	P
Maternal T2 glucose (mmol/L)a	5.41 (1.17)	BMI	0.38b	0.00591	BMI 12.5%	.011
		SAT	0.20
		VAT	0.32b
		Hip	0.28
Maternal T2 triglyceride (mmol/L)a	2.41 (0.73)	BMI	0.20	0.00348	SAT 5.1%	.078
		SAT	0.27
		VAT	0.17
		Hip	0.12
Maternal T2 HOMAa	12.3 (10.8)	BMI	0.41b	0.0313	BMI 15.1%	.005
		SAT	0.31b
		VAT	0.37b
		Hip	0.33b
Maternal AUC glucose (mmol/L × wk)a	120 (26)	BMI	0.39b	0.00563	BMI 15.3%	.009
		SAT	0.17
		VAT	0.27
		Hip	0.34b
Maternal AUC triglyceride (mmol/L × wk)	50 (14)	BMI	0.18		No modelc
		SAT	0.27
		VAT	0.06
		Hip	0.15
Maternal AUC HOMA (HOMA × wk)a	240 (186)	BMI	0.35b	0.02074	BMI 10.0%	.021
		SAT	0.25
		VAT	0.24
		Hip	0.31b

Mean (SD) maternal plasma trimester 2 levels or AUC trimester 1 to trimester 3 (n = 45) are shown below along with their univariate association (Pearson correlation) with maternal BMI, VAT, and SAT thickness and hip circumference. The relationship between maternal BMI, VAT, and upper body SAT thickness and maternal plasma metabolic measures was determined by entering BMI, SAT, VAT, and hip circumference in a stepwise regression model P to enter and P to stay .15. r² adjusted and P values are stated.

^aAnalysis carried out on log-transformed data.

^b  P < .05,

^cNo terms were at P < .15 to be entered into the model.

Abbreviations: AUC, area under the curve; BMI, body mass index; HOMA, Homeostatic Model Assessment; SAT, subcutaneous adipose tissue; T2, trimester 2; VAT, visceral adipose tissue.

Relationships between maternal plasma glucose and lipid exposure and fetal cord plasma glucose and lipids, fetal adiposity (cord plasma leptin), and birth weight centile

None of the markers of maternal glucose or lipid gestational exposure were associated with any measure of cord plasma glucose or lipid metabolism or fetal adiposity before or after accounting for mode of delivery other than maternal AUC triglyceride (r² = 20%, P =0.020, coefficient = -0.007) and maternal AUC HOMA (r² = 15%, P = .042, coefficient = 0.255), which were associated with cord plasma HDL after accounting for mode of delivery. Maternal trimester 2 HOMA (r = 0.33, P = .028) and AUC glucose (r = 0.36, P = .016) were univariately associated with birthweight centile. T2 HOMA (P = .046) and placental weight (P < .001) remained as predictors of birthweight centile in a minimal model that included these two variables and AUC glucose (r² adjusted = 31.7%).

Maternal anthropometric measures and fetal adiposity

Fetal adiposity was not associated with maternal BMI, SAT, or VAT thickness or hip circumference in multivariate analysis. However, there was a significant positive association between placental weight and cord plasma leptin (r² = 57%, P < .001). A multivariable model including maternal anthropometrics, placental weight, and mode of delivery showed that only placental weight was independently associated with fetal adiposity (r² adjusted 48%, P < .001). There was no association between cord plasma triglyceride or nonesterified fatty acids and fetal adiposity even after adjusting for mode of delivery. The 16:0/18:2 n-6 ratio of fetal erythrocyte fatty acids was used as an index of de novo lipogenesis. This index was unrelated to fetal adiposity (r = -0.13, P = .57), but inversely associated with fetal cord plasma triglyceride (r = -0.44, P = .044) (Supplemental Fig. 2 located in a digital research material depository (60)), an effect lost after accounting for mode of delivery.

Maternal LPL mass, maternal and fetal plasma glucose, and lipids, fetal adiposity, and birth weight centile

Low LPL mass is a marker of severity of metabolic syndrome and low plasma levels reflect reduced LPL synthesis by adipocytes in the insulin-resistant state. Maternal trimester 2 LPL mass was negatively correlated with maternal BMI (r = -0.36, P = .017) and SAT (r = -0.36, P = .018), whereas AUC LPL mass correlated with SAT (r = -0.30, P = .048). In a multivariate regression model including all maternal anthropometrics, maternal SAT alone predicted trimester 2 LPL mass (r² adjusted 11.3%, P = .017). Maternal trimester 2 or AUC LPL mass was not correlated with cord plasma glucose or lipid levels. Maternal trimester 2 LPL was negatively correlated with maternal trimester 2 glucose (r = -0.30, P = .049), AUC glucose (r = -0.36, P = .019), and trimester 2 HOMA (r = -0.30, P = .048). In particular, maternal trimester 2 and AUC LPL mass were strongly correlated with maternal trimester 2 triglycerides (r = -0.52, P = .001 and r = -0.55, P < .001, respectively) and AUC triglycerides (r = -0.41, P = .007 and r = -0.41, P = .006, respectively). Maternal trimester 2 and AUC LPL were not associated with birth weight centile but both were strongly associated with fetal leptin (r = -0.69, P = .001 and r = -0.58, P = .006, respectively) (Fig. 2); these associations were independent of mode of delivery.

Figure 2.

Association between fetal adiposity (cord plasma leptin) and maternal trimester 2 lipoprotein lipase mass. Univariate correlation (Pearson) between maternal trimester 2 lipoprotein lipase mass and fetal adiposity r = -0.69 P = .001 (n = 2 data missing).

Discussion

Maternal first trimester VAT thickness on ultrasound, but not first trimester BMI, abdominal SAT, or hip circumference, was independently associated with birth weight centile. This observation is in agreement with a similar study in adolescent mothers, although the previous study lacked SAT assessment (45). The strength of associations in the current study were of greater magnitude than that previously reported, possibly because of our older, more obese population. Maternal VAT was also associated with fetal cord plasma triglyceride, although the latter variable was unrelated to birth weight centile or fetal adiposity. Maternal VAT was not associated with maternal plasma lipids, as might be expected from data in nonpregnant women, in which an oversupply of fatty acids in the portal circulation to the liver can drive an increased synthesis and secretion of very low-density lipoprotein (61). This lack of relationship between maternal VAT and plasma triglyceride suggests that VAT-associated hypertriglyceridemia may be superseded by maternal metabolic adaptation to pregnancy.

We have previously shown in ex vivo adipocyte lipolysis experiments, that SAT adipocytes have higher lipolysis rates than VAT adipocytes in pregnancy (62). Thus, in healthy pregnancy, SAT rather than VAT is the primary source of the maternal fatty acids released for maternal metabolism and placental transport, secondary to pregnancy hormone-induced gestational insulin resistance (62). A slowing or reversal of maternal (subcutaneous) adipose tissue accumulation towards the end of the second trimester coincides with the accelerated phase of fetal adipose tissue accretion (14). In pregnancy, it is possible that VAT is a marker of ectopic fat accumulation rather than a direct source of lipids for transport to the fetus. In nonpregnant individuals, fat accumulation in VAT and ectopically in other organs is secondary to dysregulated adipocyte expansion in subcutaneous fat (24,63,64). Fatty acids that overspill from the SAT compartment accumulate in VAT and are stored ectopically as intracellular lipid droplets in other tissues such as the liver and pancreas. However, in pregnancy, the overspill fatty acids from SAT could also be available for uptake and transport by the placenta, thus increasing lipid supply to the fetus (Fig. 3).

Figure 3.

Proposed pathway for the contribution of maternal BMI, SAT, and VAT to fetal birth weight and adiposity. In low BMI pregnant women, SAT contains IS, hyperplasic adipocytes secreting large amounts of LPL that are capable of expanding to store excess fatty acids (NEFA). In an insulin-sensitive environment, this facilitates regulation of maternal glucose and triglyceride concentrations providing sufficient fuels for placental transport to support healthy fetal growth. In high BMI pregnant women, SAT contains IR, hypertrophic adipocytes resulting from limited expansion of preadipocytes to form mature adipocytes. SAT has a reduced ability to store fatty acid which spill over and are directed to the liver increasing plasma TG concentrations and are stored ectopically in VAT. The increasingly insulin resistance environment also raises plasma glucose levels thus increasing the supply of both fuels across a larger placenta resulting in a larger and fatter fetus. Abbreviations: BMI, body mass index; BWC, birth weight centile; IR, insulin resistant; IS, insulin sensitive; LPL, lipoprotein lipase; NEFA, nonesterified fatty acid; SAT, subcutaneous adipose tissue; TG, triglyceride; VAT, visceral adipose tissue.

Low maternal LPL mass is a measure of metabolic syndrome severity and probably reflects a reduced rate of LPL synthesis by insulin-resistant SAT adipocytes (7,65-67). In the present study, LPL mass was inversely associated with maternal plasma triglyceride and fetal adiposity, supporting the idea of a failure of SAT adipocyte expansion and the development of adipocyte insulin resistance with a consequent overspill of fatty acids and transport to the fetus (Fig. 3). Potential mechanisms for increased fetal adiposity include an increased lipid supply across the placenta or increased de novo lipogenesis from glucose supplied across the placenta. The inverse association between cord plasma triglyceride and an index of fetal de novo lipogenesis was not independent of mode of delivery, and in any case would suggest that the transported fatty acids may be used by the fetus in preference to the de novo synthesis of fatty acids. Failure of SAT adipocyte expansion has been proposed to underlie obesity-related preeclampsia (62) and gestational diabetes mellitus (68,69). In the healthy women under study here, the maximum VAT thickness measured was 27.3 mm, which may represent a propensity toward limited SAT expansion rather than a pathological state, especially when it is considered that preliminary studies indicated that preeclampsia and GDM are predicted by VAT thickness greater than 52 mm (36) and 47.4 mm (33), respectively. Assessment of VAT thickness in larger populations would be useful to assess its ability to predict similar adverse pregnancy outcomes.

Maternal placental weight had an independent association with birth weight centile in this cohort of women. Maternal body mass index was related to both placental weight and maternal insulin resistance, and through these associations may be indirectly linked to birth weight centile and fetal adiposity. Our data showed that trimester 2 markers of glucose metabolism were most strongly related to maternal anthropometrics. The middle trimester is the time of greatest acceleration in fetal growth (70), when changes in maternal metabolism would be expected to have most effect on birthweight centile. Maternal gestational insulin resistance directs more glucose for transport to the fetus. A combination of higher placental weight, increased surface area for transport of nutrients, and raised trimester 2 plasma glucose may explain the link between BMI and fetal birth weight centile and adiposity. Interestingly, maternal BMI was associated with reduced fetal cord blood insulin and HOMA, suggesting that in the present study fetuses of healthy high BMI mothers are more insulin-sensitive and hence efficient at storing fuel as adipose tissue in addition to having more insulin-induced somatic growth. However, this could be due to our small sample size and limited statistical ability to account for confounders. Previous larger studies (5,71) show a positive relationship between maternal BMI and cord blood insulin suggesting that Pedersen’s hypothesis also applies in healthy normoglycemic pregnancies. It is currently not clear to what extent fetal insulin sensitivity is directly influenced by the mother or is a fetal response to the availability of fuel.

The data presented here suggest an input from both glucose and lipids into birth weight and fetal adiposity. Although there is plentiful evidence to link maternal plasma glucose levels with fetal growth, there has been less research into the role of plasma lipids. Lipid concentrations are equally regulated by insulin and affected by insulin resistance. It is notable that a stable isotope tracer study in healthy women at 34 to 36 weeks of gestation showed that both glucose production rate and lipolysis were independently correlated with estimated fetal size on ultrasound scan (19). Our data also suggest that glucose and lipid metabolism are intertwined and influenced by both maternal adiposity and body fat distribution.

This study had a number of strengths, including its prospective design and the assessment of a number of maternal anthropometrics in parallel with measures of both maternal and fetal cord glucose and lipid metabolism. Limitations include a lack of kinetic assessment of maternal and fetal metabolites and the use of steady-state concentrations to infer such fluxes. Our conclusions with respect to fetal adiposity are limited by cord blood samples being collected in less than one-half of the cohort (23 pregnancies), in which plasma leptin concentration was measured as a surrogate. In addition, cord plasma measurements may have been confounded by a number of pregnancy factors including fetal sex, mode of delivery, gestational age, and maternal fasting status at delivery and maternal smoking, statistical adjustment for which may have been inadequate

In summary, maternal body fat distribution in healthy pregnancy, as identified by VAT thickness, provides additional information to that of maternal BMI in the prediction of birth weight centile. VAT may be acting as a marker of reduced SAT expansion, leading to increased availability of plasma fatty acids for placental transport. The data do not address the clinical value of measuring SAT and VAT over BMI because ultimately the data do not directly link either SAT or VAT to any adverse maternal or fetal pregnancy outcome. Instead the data suggest that the ability of a woman to expand her SAT depot in response to pregnancy hyperphagia may predict her metabolic response and ultimately her susceptibility to metabolic complications of pregnancy, such as preeclampsia and GDM, and her offspring’s propensity to adiposity, at least at birth. As to what is the best marker of this susceptibility (e.g., VAT depot size, plasma LPL mass) is not currently clear, but worth exploring in the future.

Abbreviations

Abbreviations
 
• AUC

  area under the time × concentration curve

 
• BMI

  body mass index

 
• GDM

  gestational diabetes mellitus

 
• HOMA

  Homeostatic Model Assessment

 
• HDL

  high-density lipoprotein

 
• LPL

  lipoprotein lipase

 
• SAT

  subcutaneous adipose tissue

 
• VAT

  visceral adipose tissue

Acknowledgments

Financial Support: Wellbeing of Women Research Training Fellowship RTF 203 (EMJ), British Medical Association Obesity Grant 2003.

Additional Information

Disclosure Summary: All authors certify that they do not have a conflict of interest that is relevant to the subject matter or materials included in this work.

Data Availability: The datasets generated during and analyzed during the current study are not publically available but are available from the corresponding author on reasonable request.