Large-for-Gestational-Age May Be Associated With Lower Fetal Insulin Sensitivity and β-Cell Function Linked to Leptin

Abstract

Context

Fetal overgrowth is associated with increased risk for type 2 diabetes in adulthood. It is unclear whether there are alterations in insulin sensitivity and β-cell function in early life.

Objective

To determine whether large-for-gestational-age (LGA) (birth weight > 90th percentile), an indicator of fetal overgrowth, is associated with altered fetal insulin sensitivity and β-cell function.

Study Design, Population, and Outcomes

In the Design, Development, and Discover birth cohort in Canada, we studied 106 pairs of LGA and optimal-for-gestational-age (OGA; birth weight, 25th to 75th percentiles) infants matched by maternal ethnicity, smoking status, and gestational age. Cord plasma glucose-to-insulin ratio was used as an indicator of fetal insulin sensitivity, and proinsulin-to-insulin ratio was used as an indicator of β-cell function. Cord plasma leptin and high-molecular-weight (HMW) adiponectin concentrations were measured.

Results

Comparisons of infants who were born LGA vs OGA, adjusted for maternal and newborn characteristics, showed that cord blood insulin, proinsulin, and leptin concentrations were significantly higher, whereas HWM adiponectin concentrations were similar. Glucose-to-insulin ratios were significantly lower (15.4 ± 28.1 vs 22.0 ± 24.9; P = 0.004), and proinsulin-to-insulin ratios significantly higher (0.73 ± 0.82 vs 0.60 ± 0.78; P = 0.005) in LGA vs OGA newborns, indicating lower insulin sensitivity and β-cell function in LGA newborns. These significant differences were almost unchanged after further adjustment for cord blood adiponectin levels but disappeared upon additional adjustment for cord blood leptin levels.

Conclusions

This study demonstrates that LGA may be associated with decreases in both fetal insulin sensitivity and β-cell function. The alterations appear to be linked to elevated leptin levels.

Children with poor or excessive fetal growth—commonly indicated by birth weight of small-for-gestational-age (<10th percentile) or large-for-gestational-age (LGA; >90th percentile)—are at increased risk of metabolic syndrome and related disorders such as type 2 diabetes in adulthood (1–3). Children who are LGA are at an increased risk for obesity (4, 5), high blood pressure (6), and insulin resistance (7, 8). The effect of such developmental “programming” in fetal growth restriction (small-for-gestational-age) may be partly attributable to reduced β-cell mass (2), whereas the programming mechanisms in excessive fetal growth (LGA) are less understood. LGA has been associated with hyperinsulinemia and elevated insulin resistance at birth in previous studies (9, 10), but the small sample sizes [22 LGA vs 10 appropriate-for-gestational-age infants (10th to 90th percentiles) and 40 LGA vs 34 appropriate-for-gestational-age infants, respectively] called for validation in larger studies. Elevated insulin resistance has been reported in prepubertal children who were born LGA (8, 11), suggesting that insulin sensitivity may be a key link in fetal overgrowth metabolic programming. We are unaware of any research data on whether LGA may affect fetal β-cell function. In a large birth cohort, we sought to assess the impact of LGA on fetal insulin sensitivity and β-cell function.

Leptin and adiponectin are important adipokines in the regulation of insulin sensitivity (12) and have been linked to fetal growth (13). A negative correlation has been observed between cord blood leptin and fetal insulin sensitivity in our previous study (14) and other studies (15), suggesting that leptin may be involved in glucose metabolic health in early life. High-molecular-weight (HMW) adiponectin is the predominant form of adiponectin in cord blood (16, 17) and the bioactive form of adiponectin in term of its insulin sensitizing property (18, 19). However, previous studies did not detect an association between total or HMW adiponectin and fetal insulin sensitivity (20, 21). There is a lack of data on whether leptin and adiponectin are implicated in any changes in fetal insulin sensitivity in infants who were born LGA. In the current study, we sought to assess whether leptin and HMW adiponectin may be involved in any alterations in fetal insulin sensitivity or β-cell function in infants who were born LGA.

Materials and Methods

Study design

We conducted a matched case-control study nested in the recently reported Design, Development, and Discover (3D) birth cohort (n = 2366) developed by the Integrated Research Network of Perinatology in Quebec and Eastern Ontario (22). In the 3D cohort, women bearing a singleton fetus were recruited at 8 to 14 weeks of gestation between May 2010 and August 2012 in nine obstetric care centers in Quebec, Canada. The study was approved by the research ethics committees of Sainte-Justine University Hospital Research Center (the coordination center, research ethics approval number 2899) and all participating hospitals. Written informed consent was obtained from all study participants.

LGA was defined as birth weight exceeding the 90th percentile for gestational age, according to the Canadian sex- and gestational age–specific birth weight standards (23). There were a total of 165 infants who were born LGA at delivery. We excluded infants who were born LGA and conceived with the use of artificial reproductive technology (n = 20) or with any known birth defect (n = 1), or infants who were born LGA with missing cord blood specimen or specimens with evidence of hemolysis (n = 38). This left 106 infants who were born LGA (gestational age at delivery, 33 to 41 weeks) eligible in the current study. For each infant who was LGA, we randomly selected a matched optimal-for-gestational-age (OGA; birth weight, 25th to 75th percentiles) control infant. Eligible controls were required to have a cord blood specimen available without evidence of hemolysis. The matching factors were ethnicity (white, others), smoking status (current smokers, previous smokers, nonsmokers), and gestational age at delivery (moderate preterm, 33 to 36 weeks; term ≥ 37 weeks). The flowchart in the selection of study participants is presented in Fig. 1.

Figure 1.

Study flowchart in the selection of infants who were born LGA (birth weight > 90th percentile) and matched OGA (birth weight, 25th to 75th percentiles) control infants in the 3D birth cohort.

Specimens and biochemical assays

In the 3D cohort, research assistants were available 24 hours a day on-call for timely collection and processing of cord blood specimens. All specimens were kept on ice and centrifuged within 30 minutes after collection. Plasma samples were stored in multiple aliquots at −80°C until biochemical assays.

Cord plasma glucose concentrations (in mmol/l; 1 mmol/L = 18 mg/dL) were determined by an automated glucose oxidase method (Beckman-Coulter, Brea, California). Cord plasma insulin (in pmol/L; 1 mU/L = 6 pmol/L) was measured by an automated ultrasensitive chemiluminescent immunometric assay (Beckman-Coulter), proinsulin by a quantitative ELISA kit (ALPCO Diagnostics, Salem, NH), leptin by a human leptin ELISA kit (Invitrogen, Camarillo, CA), and HMW adiponectin by a human HMW adiponectin ELISA kit (MyBioSource, San Diego, CA). The interassay and intra-assay coefficients of variation of these assays were in the range of 2.0% to 9.6%.

Outcomes

The primary outcomes were cord plasma glucose-to-insulin ratio as an indicator of fetal insulin sensitivity (24, 25) and proinsulin-to-insulin ratio as an indicator of fetal β-cell function (26, 27). More accurate methods in estimating insulin sensitivity and β-cell function (such as the euglycemic clamp) are impractical at birth. Other outcomes included cord plasma insulin and proinsulin concentrations.

Statistical analysis

Continuous variables are presented as mean ± SD or median (interquartile range). Except for glucose (normal distribution), data for biomarkers (positively skewed data distribution) were log-transformed in paired t tests for differences between LGA and OGA groups and in generalized linear models to assess the differences adjusting for maternal and infant characteristics. These characteristics included maternal age (<35 years and ≥35years), ethnicity [white (the majority) and others], education (university: yes/no), prenatal smoking (current, previous, no), alcohol consumption (current, previous, no), parity (primiparous: yes/no), mode of delivery (cesarean, vaginal), gestational diabetes mellitus (yes/no), gestational hypertension and preeclampsia (yes/no), prepregnancy obesity (body mass index ≥ 30.0 kg/m², yes/no], duration of labor (hour), infant sex, Apgar score, umbilical artery pH, and gestational age. To obtain more stable estimates of the differences between infants who were born LGA and OGA, only covariables associated with LGA (significantly different between the two groups) or affecting the comparisons (≥10% changes in regression coefficients) were included in the final adjusted models. All data management and analyses were conducted using SAS software, version 9.2 (SAS Institute, Cary, NC). A P value < 0.025 was considered to indicate a statistically significant difference, accounting for multiple tests (two primary outcomes of interest).

Results

Maternal and infant characteristics

A comparison of infants who were born LGA and OGA showed no significant differences in maternal ethnicity, education status, tobacco smoking, alcohol use, prepregnancy body mass index and obesity, gestational diabetes mellitus, gestational hypertension and preeclampsia, or preterm delivery (Table 1). Comparing LGA vs OGA groups, there were more mothers with advanced maternal age (≥35 years: 22.6% vs 11.3%), and more cesarean deliveries (41.5% vs 17.0%), but fewer primiparous women (34.0% vs 49.1%). Seventeen pregnancies were complicated by gestational diabetes mellitus according to the American Diabetes Association’s 2-hour 75-g oral glucose tolerance test diagnostic criteria (28), including 10 (9.4%) in the LGA group and 7 (6.6%) in the OGA group. Twelve women (six in each group) delivered preterm (all mildly preterm, at 33 to 36 weeks). As expected, birth weights were substantially higher in infants who were born LGA than in infants who were born OGA (4244.2 ± 306.1 g vs 3365.2 ± 277.2 g). Low Apgar score (<7) was rare (only four cases). There was no evidence of acidosis (cord artery blood pH < 7.05) in any neonate.

Table 1.

Maternal and Infant Characteristics in Infants Born LGA and OGA

Characteristic	LGA (n = 106)	OGA (n = 106)	P Value a
Maternal
Age, y	31.7 ± 4.8	30.6 ± 4.1	0.06
≥35 y	24 (22.6)	12 (11.3)	0.03
White ethnicity	75 (70.8)	75 (70.8)	1.0
Primiparity	36 (34.0)	52 (49.1)	0.03
University education	82 (58.6)	89 (65.0)	0.27
Tobacco smoking			1.0
None	74 (69.8)	74 (69.8)
Previous	19 (17.9)	19 (17.9)
Current	13 (12.3)	13 (12.3)
Alcohol consumption (missing: n = 17)			0.25
None	56 (57.1)	45 (46.4)
Previous	35 (25.7)	46 (47.4)
Current	7 (7.2)	6 (6.2)
Prepregnancy BMI, kg/m2	25.8 ± 5.6	24.3 ± 5.7	0.05
Obesity (BMI ≥ 30.0 kg/m2)	19 (18.5)	17 (16.4)	0.69
Gestational diabetes	10 (9.4)	7 (6.6)	0.45
Maternal fasting glucose, mmol/L	4.6 ± 0.44	4.4 ± 0.44	0.17
Gestational hypertension	8 (7.6)	9 (8.5)	0.80
Preeclampsia	1 (0.9)	3 (2.8)	0.31
Cesarean delivery	44 (41.5)	18 (17.0)	<0.001
Duration of labor, h	7.7 ± 9.0 (n = 86)	8.4 ± 8.5 (n = 75)	0.60
Infants
Birth weight, g	4244.2 ± 306.1	3365.2 ± 277.2	<0.001
Preterm birth (<37 wk)	6 (5.7)	6 (5.7)	1.0
Male	48 (45.3)	46 (43.4)	0.78
Apgar score < 7	0 (0)	4 (3.8)	0.06
Umbilical artery pH	7.26 ± 0.06	7.26 ± 0.07	0.99
Acidosis (pH < 7.05)	0 (0)	0 (0)

Data presented are mean ± SD or n (%). Infants who were LGA had birth weight > 90th percentile, according to the Canadian sex- and gestational age–specific birth weight standards; infants who were born OGA had birth weight in the 25th to 75th percentiles.

^a

P values from χ² tests for differences in categorical variables or paired t tests for differences in continuous variables.

Correlations

There were no significant interactions (all P > 0.1) between LGA status and the associations of cord blood leptin or HMW adiponectin with glucose-to-insulin ratio or proinsulin-to-insulin ratio. We thus presented the correlations in the total study sample (rather than in infants who were born LGA and OGA separately). Cord plasma leptin was positively correlated to birth weight (r = 0.56; P < 0.001), cord plasma insulin (r = 0.40; P < 0.001), proinsulin (r = 0.56; P < 0.001), and proinsulin-to-insulin ratio (r = 0.20; P = 0.003) and negatively correlated to glucose-to-insulin ratio (r = −0.41; P < 0.001) (Table 2). Cord blood HMW adiponectin was positively correlated to glucose-to-insulin ratio (r = 0.26; P < 0.001), was not significantly correlated to proinsulin-to-insulin ratio (r = 0.14; P = 0.05), and was not correlated with birth weight (P = 0.41).

Table 2.

Partial Correlations of Leptin and HMW Adiponectin With Birth Weight and Metabolic Health Biomarkers in Cord Blood

Variable	Leptin		HMW Adiponectin
	γ	P Value	γ	P Value
Birth weight (z score)	0.56	<0.001	−0.06	0.41
Glucose	−0.04	0.57	0.10	0.14
Insulin	0.40	<0.001	−0.23	<0.001
Proinsulin	0.56	<0.001	−0.13	0.07
Glucose-to-insulin ratio	−0.41	<0.001	0.26	<0.001
Proinsulin-to-insulin ratio	0.20	0.003	0.14	0.05

Partial correlation coefficients adjusted for gestational age at delivery; correlation coefficients with P < 0.004 are significant after accounting for multiple correlation tests.

Differences between infants who were born LGA and OGA

Comparison of infants who were born LGA vs OGA adjusted for maternal and newborn characteristics showed cord blood glucose levels were similar, whereas insulin and proinsulin levels were significantly higher in infants who were born LGA (mean ± SD, 74.9 ± 81.8 pmol/L and 44.8 ± 63.9 pmol/L, respectively) than in infants who were born OGA (40.3 ± 32.1 pmol/L and 17.6 ± 16.7 pmol/L, respectively) (all adjusted P < 0.004). Cord blood leptin levels were significantly higher (25.8 ± 20.1 ng/mL vs 13.2 ± 9.5 ng/mL; adjusted P < 0.001), whereas glucose-to-insulin ratios were significantly lower (15.4 ± 28.1 vs 22.0 ± 24.9; adjusted P = 0.004) and proinsulin-to-insulin ratios were significantly higher (0.73 ± 0.82 vs 0.60 ± 0.78; adjusted P = 0.005) in infants who were born LGA vs OGA (Table 3 and Fig. 2). Cord plasma HWM adiponectin concentrations were similar between infants who were born LGA and OGA.

Table 3.

Cord Blood Metabolic Biomarkers in Infants Who Were LGA vs OGA

Variable	LGA (n = 106)	OBW (n = 106)	Crude P Value a	Adjusted P Value a
Glucose, mmol/L			0.33	0.25
Median (Q25, Q75)	4.3 (3.5, 5.1)	4.5 (3.8, 5.1)
Mean ± SD	4.4 ± 1.1	4.5 ± 1.0
Insulin, pmol/L			<0.001	0.003
Median (Q25, Q75)	52.5 (32.6, 87.9)	33.2 (19.0, 51.2)
Mean ± SD	74.9 ± 81.8	40.3 ± 32.1
Proinsulin, pmol/L			<0.001	<0.001
Median (Q25, Q75)	22.3 (14.5, 39.1)	13.2 (9.4, 22.6)
Mean ± SD	44.8 ± 63.9	17.6 ± 16.7
Glucose-to-insulin ratio			<0.001	0.004
Median (Q25, Q75)	8.6 (5.3, 14.5)	15.0 (8.8, 24.8)
Mean ± SD	15.4 ± 28.1	22.0 ± 24.9
Proinsulin-to-insulin ratio			0.04	0.005
Median (Q25, Q75)	0.48 (0.34, 0.67)	0.43 (0.29, 0.68)
Mean ± SD	0.73 ± 0.82	0.60 ± 0.78
Leptin, ng/mL			<0.001	<0.001
Median (Q25, Q75)	18.4 (11.0, 30.9)	10.6 (7.4, 15.4)
Mean ± SD	25.8 ± 20.1	13.2 ± 9.5
HMW adiponectin, µg/mL			0.14	0.58
Median (Q25, Q75)	19.5 (13.8, 25.0)	21.0 (15.0, 27.1)
Mean ± SD	20.1 ± 9.4	21.5 ± 9.3

Infants who were born LGA had birth weight > 90th percentile, according to the Canadian sex- and gestational age–specific birth weight standards; infants who were born OGA had birth weight in the 25th to 75th percentiles. Q25, 25th percentile; Q75, 75th percentile.

^a

P values in comparisons of infants who were born LGA vs infants who were born OGA in log-transformed biomarker data. Crude P values were from paired t tests. Adjusted P values were from generalized linear models adjusted for maternal age, parity, and mode of delivery; other factors were not adjusted for because they were similar and did not affect the comparisons between the two groups (although the fully adjusted results were similar).

Figure 2.

Comparisons of cord blood biomarkers in infants who were born LGA (birth weight > 90th percentile) and OGA (birth weight, 25th to 75th percentiles). The error bars represent the 95% CIs of mean values (crude data). The P values were from comparisons in log-transformed data adjusted for maternal age, parity, and mode of delivery; other factors were not adjusted for because they were similar and did not affect the comparisons.

Sensitivity analyses

Results were similar when the analyses were restricted to full-term, nondiabetic, and nonpreeclampsia pregnancies (n = 91 LGA-control pairs). For example, cord plasma insulin and proinsulin levels remained significantly higher in infants who were born LGA (65.7 ± 66.7 pmol/L and 38.2 ± 49.2 pmol/L, respectively) than in infants who were born OGA (37.9 ± 28.4 pmol/L and 17.5 ± 17.7 pmol/L, respectively) (all P ≤ 0.001), whereas glucose-to-insulin ratios remained significantly lower (16.9 ± 30.0 vs 22.0 ± 23.4; P = 0.008) and proinsulin-to-insulin ratios remained significantly higher (0.77 ± 0.87 vs 0.58 ± 0.66; P = 0.02) in infants who were born LGA.

Stepwise adjusted differences

After adjustment for maternal age, parity, and mode of delivery (other covariables did not affect the comparisons), the differences between infants who were born LGA and OGA were attenuated but remained statistically significant for cord blood insulin (44.6% higher; P = 0.003), proinsulin (90.1% higher; P < 0.004), and glucose-to-insulin ratio (28.6% lower; P = 0.004) but widened for proinsulin-to-insulin ratio (32.9% higher; P = 0.005) (Table 4, model 1). These significant differences remained almost unchanged with further adjustment for cord blood HMW adiponectin levels (model 2) but disappeared with further adjustment for cord blood leptin levels (model 3).

Table 4.

Percentage Differences in Cord Plasma Insulin, Proinsulin, Glucose-to-Insulin Ratio, and Proinsulin-to-Insulin Ratio in LGA vs OGA

Variable	Crude		Model 1		Model 2		Model 3
	% Change (95% CI)	P Value	% Change (95% CI)	P Value	% Change (95% CI)	P Value	% Change (95% CI)	P Value
Insulin	66.0 (32.0–108.9)	<0.001	44.6 (14.0–83.3)	0.003	43.8 (13.9–81.6)	0.002	9.9 (−13.3 to 39.1)	0.43
Proinsulin	99.8 (60.5–148.8)	<0.001	90.1 (50.7–139.8)	0.004	89.8 (50.4–139.7)	<0.001	32.0 (5.9– 64.5)	0.01
Glucose-to-insulin ratio	−42.1 (−53.9 21 to −27.2)	<0.001	−28.6 (−43.1 to −10.6)	0.004	−28.2 (−42.6 to −10.3)	0.004	−6.7 (−25.5 to 16.9)	0.55
Proinsulin-to-insulin ratio	21.5 (0.5, 46.9)	0.04	32.9 (9.1, 61.8)	0.005	32.0 (8.5, 60.5)	0.006	20.2 (−2.7 to 48.4)	0.09

Iinfants who were born LGA had birth weight > 90th percentile, according to the Canadian sex- and gestational age–specific birth weight standards; infants who were born OGA had birth weight in the 25th to 75th percentiles. The percentage change was calculated from the regression coefficient of the dependent variable (y) in log scale; because the regression coefficient (β) represents the proportion of change in y in the original scale: log y₁ – log y₀ = β, then log (y₁/y₀) = β. Thus, y₁/y₀=e^β, and thus the percentage change is (e^β −1) ×100%. Model 1: adjusted for maternal age, parity, and mode of delivery mode; other maternal and infant characteristics were not adjusted for because they were similar and did not affect the comparisons between the two groups. Model 2: model 1 plus further adjustment for cord blood HMW adiponectin levels. Model 3: model 2 plus further adjustment for cord blood leptin levels.

Discussion

Main findings

We observed that infants who were born LGA had lower cord plasma glucose-to-insulin ratios, indicating lower fetal insulin sensitivity, and higher proinsulin-to-insulin ratios, indicating lower β-cell function relative to infants who were born OGA. Moreover, this was true even after exclusion of gestational diabetes mellitus—a major pregnancy complication suspected to be blamed for “pathological” LGA. The adjusted analyses suggest that these changes were related to higher cord blood leptin levels but were not related to HMW adiponectin levels. This study detected a negative association between cord blood leptin and fetal β-cell function.

Data interpretation and comparisons to previous studies

Studies of metabolic health biomarkers in cord blood or neonatal blood soon after birth have reported positive associations between fetal overgrowth and insulin resistance (9, 29, 30). This was confirmed in our study; LGA was associated with lower fetal insulin sensitivity, as indicated by lower cord plasma glucose-to-insulin ratios. In this largely healthy group of newborns (only four, all in the OGA group, had an Apgar score < 7; all infants had cord artery pH > 7.05), we detected clear differences in cord blood glucose metabolic health biomarkers between infants who were born LGA and OGA independent of neonatal characteristics.

The fetal pancreas is relatively immature, which may explain the substantially higher proinsulin-to-insulin ratios in cord blood than in adult blood (31, 32). We observed that infants who were born LGA had higher proinsulin-to-insulin ratios (despite similar glucose levels) indicating lower β-cell capacity in converting proinsulin to insulin. Elevated proinsulin-to-insulin ratios indicate β-cell stress (33). Our results suggest that fetal overgrowth may be linked to poorer pancreatic β-cell function. We speculate that in LGA fetuses/neonates, there may be some fetal adaptations in response to an adverse intrauterine environment affecting pancreatic β cells.

Our data confirmed the higher cord blood leptin levels in infants who were born LGA, the positive correlation between cord blood leptin and birth weight (34, 35), and the negative association between cord blood leptin and fetal insulin sensitivity (14, 36). Furthermore, we observed a negative association of fetal leptin with β-cell function (as indicated by the positive correlation with proinsulin-to-insulin ratio). The adjusted analyses suggest that the lower fetal insulin sensitivity (higher glucose-to-insulin ratios) and lower β-cell function (higher proinsulin-to-insulin ratios) in infants who were born LGA were linked to higher leptin levels. However, the current study cannot determine whether the latter is a cause, consequence, or accomplice of the former. The results suggest the possible involvement of leptin in the regulation of fetal insulin sensitivity and β-cell function, raising the possibility that leptin may play an important role in metabolic health homeostasis during fetal life.

Adiponectin is an endogenous insulin sensitizer and is richly expressed in fetal life predominantly in the form of HMW adiponectin (16, 17). However, previous studies did not detect an association between cord blood total or HMW adiponectin and fetal insulin sensitivity (14, 20, 21). In the current study, we detected a positive correlation between HMW adiponectin and fetal insulin sensitivity. The reasons behind the discrepant findings are unclear and may be partly due to the differences in sample size (ours is relatively larger). More studies are needed to understand the implications of fetal adiponectin level for both fetal and postnatal metabolic health.

In contrast to decreased insulin sensitivity and β-cell function but increased leptin levels in infants who were born LGA, increased insulin sensitivity but decreased insulin secretion and leptin levels have been reported in neonates with poor fetal growth (37–39). It appears that infants at the two opposite extremes of birth weight experienced vastly divergent metabolic “programming” in utero, yet both are at increased risk for insulin resistance–related disorders in adulthood (1–3). The human biology appears to be in favor of moderation, rather than extremity at both ends. Longitudinal studies into infancy and childhood are required to illuminate the pathways into the metabolic programming effects at both extremities.

Our findings may have potential clinical implications in the care of infants who were born LGA. Considering the lower insulin sensitivity and β-cell function linked to higher leptin levels in LGA neonates, there may be a need for consideration of postnatal checkups on glucose metabolic health indicators in infants who were born LGA. High cord blood leptin levels may indicate decreased insulin sensitivity and β-cell function in LGA neonates. It should be cautioned that before further considerations on clinical implications and utility, confirmative data on the study findings from other independent cohorts are required, and further studies are required on the long-term metabolic health implications of these alterations observed at birth.

Strengths and limitations

Main strengths are the relatively large sample size, the timely collection and processing of cord blood specimens, and high-quality biomarker assays. The main limitation is that we used relatively imprecise surrogate biomarker indicators of fetal insulin sensitivity and β-cell function because the more accurate standard protocols for assessing insulin sensitivity (such as the euglycemic insulin clamp) are impractical to implement at birth. However, such imprecisions would only tend to increase random noise variations and would tend to bias the associations toward the null. Our study was based on a Canadian cohort of largely white participants. We could not meaningfully analyze the effects in other ethnic groups because of small numbers. Further studies in other countries/populations are required to understand the generalizability of the study findings.

In summary, LGA was associated with decreased fetal insulin sensitivity and β-cell function. These alterations appear to be linked to elevated fetal leptin levels.

Abbreviations:

Abbreviations:
 
• 3D

  Design, Development, and Discover

 
• HMW

  high-molecular-weight

 
• LGA

  large-for-gestational-age

 
• OGA

  optimal-for-gestational-age

Acknowledgments

This work was based on the 3D cohort developed by the Integrated Research Network in Perinatology of Quebec and Eastern Ontario.

Financial Support: This work was supported by research grants from the Canadian Institutes of Health Research (CIHR grants 88413,151517, and 158616; W.D.F., Z.-C.L.); the National Natural Science Foundation of China (grant 81571451; Z.-C.L.). W.D.F. holds a CIHR Tier 1 Canada Research Chair in Perinatal Epidemiology.

Author Contributions: Y.D., Z.-C.L., A.M.N., F.A., S.-Q.W., H.A.A., E.B., P.J., H.H., E.L., and W.D.F. conceptualized the study; Z.-C.L., A.M.N., F.A., S.-Q.W., H.A.A., E.B., P.J., E.L., and W.D.F obtained the research funding; Z.-C.L., A.M.N., F.A., S.-Q.W., H.A.A., E.B., P.J., E.L., and W.D.F. contributed to the acquisition of research data; Y.D. and Z.-C.L. conducted the data analysis; Y.D. drafted the manuscript; all authors provided critical revisions of the manuscript for important intellectual content and approved the final version for publication; Z.-C.L. has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Disclosure Summary: The authors have nothing to disclose.