Higher Offspring Birth Weight Predicts the Metabolic Syndrome in Mothers but Not Fathers 8 Years After Delivery: The Pune Children’s Study

Details of our methods have been reported previously (16). Of 477 families, 6 mothers and 12 fathers had died, and 459 (nonpregnant) mothers and 398 fathers agreed to be studied. Standardized anthropometric measurements were made by one of two trained observers. Fat mass was calculated from four skinfold-thickness measurements using Durnin’s formula (17). Blood pressure was measured using a digital automated device (Dinamap; Criticon, Tampa, FL) after 10-min rest in the supine position.

A 75-g anhydrous oral glucose tolerance test (OGTT) was performed on subjects not on antidiabetic treatment. Three venous blood samples (fasting and at 30 and 120 min after oral glucose) were collected for measurement of plasma glucose and insulin concentrations. Plasma cholesterol, HDL cholesterol, triglycerides, proinsulin, and 32-33 split proinsulin were measured in the fasting sample by standard methods. Insulin resistance was calculated using the homeostasis model assessment (HOMA) (18), and the 30-min insulin increment was used as an estimate of β-cell function (19). Metabolic syndrome was defined following World Health Organization guidelines (20), except that the BMI cutoff point for obesity was 25 kg/m² rather than 30 kg/m² because of the recent recognition of a need to revise BMI criteria for Asian populations (21). Those in the highest quartile of HOMA for insulin resistance (HOMA-IR) were diagnosed as insulin resistant. Metabolic syndrome was diagnosed if a subject had a fasting plasma glucose concentration ≥6.1 mmol/l or a 120-min concentration ≥7.8 mmol/l and/or were insulin resistant and in addition had two or more of the following: 1) BMI ≥25 kg/m² and/or waist-hip ratio ≥0.90 for men and ≥0.85 for women, 2) plasma triglyceride concentration ≥1.7 mmol/l and/or HDL cholesterol <0.9 mmol/l for men and <1.0 mmol/l for women, and 3) blood pressure ≥140/90 mmHg.

Socioeconomic status (SES) was assessed using the Kuppuswamy score (22), a standardized scoring system based on a questionnaire asking for details of occupation, education, income, and size and type of housing. Each aspect is separately scored, and the scores are added to get a total socioeconomic score (a high value indicates prosperity). History of smoking and alcohol consumption was recorded as never, past, and current.

Variables not normally distributed were log transformed to satisfy assumptions of normality and are represented in the tables by the geometric mean. For convenience of interpretation, we have shown the distribution of parental characteristics by sex-specific thirds of their child’s birth weight. However, in the analyses, birth weight and parental measurements were used as continuous variables. The significance of the relation between the child’s birth weight and parental characteristics was tested by linear regression analysis, adjusting for possible confounders (child’s sex and birth order, parental age, SES, and BMI). The significance of the relation between the child’s birth weight and categorical outcomes (obesity, diabetes, hypertension, metabolic syndrome, etc.) was tested using logistic regression.

The mean birth weights of the children were 2.8 kg (boys) and 2.6 kg (girls). Their birth weight distribution was similar to that for all babies born in the hospital during the same time period. There was no significant difference in birth weight of children whose parents were studied or not studied. Birth weight increased with the birth order of the child (P < 0.001), but was not related to the SES of the family (P = 0.42).

Table 1 shows the characteristics of the parents. Tables 2 and 3 show maternal and paternal size and cardiovascular risk factors by thirds of their child’s birth weight.

One mother was known to be diabetic before pregnancy and died 5 years after delivery. Her child and his father are excluded from the analysis.

The OGTT was not routinely performed during pregnancy, but 13 mothers were diagnosed as having GDM by the treating obstetrician on clinical indications; none of them were on antidiabetic treatment at the time of this study. OGTTs were performed in 456 mothers, of whom 16 showed diabetes (6 were previously diagnosed GDM) and 45 showed impaired glucose tolerance (IGT).

Anthropometric measurements were available in 459 mothers. Altogether, 27% had BMI >25 kg/m², 10% had waist-to-hip ratio (WHR) >0.85, and 42% had body fat >35% (Table 2). All indices of obesity (BMI, waist circumference, WHR, and body fat mass and percentage) were positively related to circulating concentrations of glucose and insulin (fasting and 30 and 120 min), triglycerides, and total cholesterol and inversely related to HDL cholesterol. HOMA-IR was strongly related to all the indices of obesity (r = ∼0.4, all P < 0.001), but relatively weakly related to head circumference (r = 0.15, P < 0.01) and not related to height (r = 0.04, P = 0.33). The prevalence of diabetes, hypertension, and hypertriglyceridemia was directly related to BMI (all P < 0.001). Higher SES was positively related to all anthropometric parameters in mothers (P < 0.001) and to plasma cholesterol concentration (P < 0.05), but not related to other cardiovascular risk factors. None of the mothers smoked or consumed alcohol.

There was a positive association between the birth weight of the child and maternal height, weight, BMI, obesity (BMI >25 kg/m²), waist circumference, and fat mass 8 years later (Table 2). Offspring birth weight was not significantly related to maternal plasma glucose concentrations after fasting and after oral glucose load. The relative risk ratio for diabetes for mothers who gave birth to larger or smaller babies compared with those in the middle was not significant [2.61 (CI 0.71–9.80) and 1.68 (CI 0.41–6.89), respectively]; similarly, prevalence of IGT was not related to child’s birth weight.

Maternal fasting plasma insulin concentration (P < 0.01) and HOMA-IR (P < 0.01) were directly related to offspring birth weight (adjusted for the child’s sex and birth order and maternal age) (Table 2). This was true even when the analysis was limited to mothers with normal glucose tolerance (P < 0.01). These relationships remained significant after adjusting for socioeconomic status but became nonsignificant after adjusting for maternal BMI. Fasting plasma proinsulin and 32-33 split proinsulin concentrations and 30- and 120-min plasma insulin concentrations were not related to offspring birth weight. There was no significant relation between offspring birth weight and maternal blood pressure or plasma cholesterol concentrations (total, LDL, and HDL), but plasma triglyceride concentration was directly related (P < 0.05). Offspring birth weight was a strong predictor of metabolic syndrome in the mother (P < 0.01).

Anthropometric measurements were available for 395 fathers and OGTT results for 398 fathers (Table 1). Altogether, 27% had BMI >25 kg/m², 57% had WHR >0.90, and 52% had body fat >25%. Plasma glucose and insulin (fasting and 30- and 120-min), HOMA-IR, triglyceride, and cholesterol concentrations were directly related (P < 0.001) and HDL cholesterol concentrations inversely related (P < 0.05) to BMI. The prevalence of diabetes, hypertension, and hypertriglyceridemia was directly related to BMI (P < 0.001). Higher SES was associated with larger size (height, BMI, waist circumference, and fat mass, all P < 0.001), higher plasma glucose (fasting and 120-min, P < 0.05), cholesterol and triglyceride concentrations (both P < 0.01), and diastolic blood pressure (P < 0.05).

Offspring birth weight was directly related to paternal height, weight, head and waist circumferences, and fat mass (Table 3). There were no significant relationships between offspring birth weight and paternal biochemical measurements (circulating glucose and insulin concentrations and HOMA-IR), blood pressure, or morbidity measures (Table 3). The birth weight of offspring of diabetic fathers (2.71 vs. 2.77 kg; P = 0.50), insulin-resistant fathers (2.81 vs. 2.76 kg; P = 0.41), or those with the metabolic syndrome (2.77 vs. 2.76 kg; P = 0.84) did not differ significantly from the remainder, even after excluding families with hyperglycemic mothers.

Paternal and maternal height (r = 0.29, P < 0.001) and BMI (r = 0.24, P < 0.001) were significantly correlated. Paternal size was not related to maternal HOMA-IR after adjusting for maternal size. Maternal body size and HOMA-IR relations with offspring birth weight remained significant after adjusting for paternal height or BMI. After adjusting for maternal height or BMI, paternal height and weight, but not waist circumference or fat mass, remained significantly related to offspring birth weight. Adjustment for maternal BMI did not reveal any significant relationship between offspring birth weight and paternal metabolic-endocrine measures, including HOMA-IR.

Our study showed significantly increased obesity, insulin resistance, hypertriglyceridemia, and prevalence of metabolic syndrome in urban Indian mothers 8 years after the birth of heavier babies. There was no relation between offspring birth weight and maternal plasma glucose concentrations (fasting and post–glucose challenge) either including or excluding diabetic mothers. Our results suggest that the persistent metabolic abnormality in the urban Indian mothers 8 years after the birth of a heavier child is insulin resistance with compensated β-cell function rather than hyperglycemia. The mean birth weight of babies in the highest one-third was only 3.3 kg. These relationships were independent of the child’s sex and maternal age, parity, and SES, but not maternal BMI, suggesting that determinants of body size (genetic, nutritional, and metabolic-endocrine) may be underlying factors. Obesity and insulin resistance are strong risk factors for diabetes and coronary heart disease (CHD). We therefore speculate that in the future there will be higher rates of diabetes and CHD in mothers of heavier babies. Though the child’s birth weight was related to paternal size 8 years later, there were no relations with paternal glucose tolerance, insulin resistance, or other cardiovascular risk factors.

Correlations between offspring birth weight and maternal size are expected. Mothers of heavy babies were fatter, taller, and had larger head circumference. This could reflect better nutrition and growth throughout life and/or genes for larger size. SES probably contributes to this relationship, although it remained significant after adjusting for SES.

It is general knowledge that mothers who give birth to heavy babies have an increased risk of diabetes, but there is little information on maternal risk of metabolic syndrome and cardiovascular risk. Serial studies of maternal metabolism before and during pregnancy have shown a relation between both pregestational (“chronic”) and late gestational maternal insulin resistance and offspring birth weight (23) and with diabetes in the mother within 2 years of delivery (24,25). The increased size of the baby is thought to be due to maternal hyperglycemia during pregnancy causing fetal hyperinsulinemia and promoting insulin-mediated growth (26). In our study, offspring birth weight predicted maternal insulin resistance 8 years after delivery, which was independent of age, parity, and SES and was also present in the normal glucose-tolerant mothers (at the time of testing, excluding those with a history of GDM). There was no relationship between offspring birth weight and maternal glycemia 8 years after delivery. One interpretation of our findings is that insulin resistance may be on the causal pathway in the maternal-offspring size relationship. The likely sequence of events is that mothers of heavier babies were more obese at the time of pregnancy (and therefore more insulin resistant) and remained so 8 years later. Our findings indirectly support the possibility that maternal circulating nonglucose nutrients (amino acids, triglycerides, fatty acids, etc.) may be important determinants of fetal size (27) and that maternal insulin resistance helps facilitate transfer to the fetus. Our data are consistent with the concept that maternal size and metabolic status, even in the absence of gestational diabetes, may have a major influence on offspring size at birth.

An alternative explanation for the birth weight–insulin resistance relationship would be “reverse causality.” A large fetus may promote increased maternal insulin resistance through the secretion of larger quantities of placental hormones (lactogen, growth hormone, etc.) (28). This could occur if the fetus is “genetically” large (paternal influence). However, we did not find a relation between paternal height or BMI and maternal HOMA-IR, independent of maternal size. A reverse causality explanation could hasten maternal β-cell decompensation, and thus later diabetes, but does not readily explain either persistent obesity or insulin resistance in the mother 8 years after delivery.

Our findings are different from those reported by Lawlor et al. (9), who found an inverse relation between offspring birth weight and maternal HOMA-IR in elderly British women (69 years of age). There may be a number of explanations for this difference. Causes of low birth weight may be different in the two populations; in developed countries, these include prematurity, pregnancy-induced hypertension, and maternal smoking, which would increase the risk of future cardiovascular disease in the mother (29,30). On the other hand, low birth weight in India is related to the small size of mothers and poor nutrition (31–33), which would protect against obesity, insulin resistance, and CHD. Also, Indian babies (34,35) and adults (36,37) have a higher percentage body fat for a given BMI compared with Caucasians. This may alter the relation between body size and disease risk in Indians compared with Caucasians (38,39). The relation between offspring birth weight and maternal insulin resistance may change with time from delivery (9,10), although “survivor bias” may contribute to the changing relationship. For example, insulin-resistant and obese mothers with polycystic ovarian disease will be selectively excluded from a retrospective cross-sectional cohort because of infertility or early pregnancy loss or because of premature death due to diabetes or heart disease (40). Finally, the differences in the socioeconomic and nutritional transition may explain the difference: in Europe, obesity, diabetes, and CHD are more common in the lower social classes who have lower–birth weight babies, but in India, these conditions are more common in the affluent, who produce heavier babies.

The independent relation between paternal height and offspring birth weight is consistent with the idea that the paternal genome has a strong influence on fetal skeletal growth (41); however, other size relationships (measures of obesity) may reflect an “assortive mating” effect because they did not remain significant after adjusting for maternal size. Lack of relation between offspring birth weight and paternal insulin resistance or glucose tolerance is unlike the findings in Pima Indians (42) and in the U.K. (10), both of which showed an inverse relationship. Thus, our data does not support the paternal predictions of fetal insulin hypothesis (5). This may be partly due to lower prevalence of diabetes in our population than among the Pima Indians. However, in a rural Indian population, paternal insulin resistance was directly related to offspring birth weight, a relationship that was dependent on paternal BMI (43,44). Paternal genetic influences may control fetal size rather than metabolism.

Barker (1) has proposed that the risk of diabetes and CHD will diminish over time as maternal nutrition, and thus fetal growth, improves. CHD has shown a downward trend in some Western countries (at least partly contributed to by improved lifestyle) but not diabetes. “Improvements” in the nutritional status of Indian mothers that simply increase their adiposity may increase fetal size and improve perinatal outcomes, but could carry a price in terms of an increased risk of diabetes and the metabolic syndrome in both mother and offspring. The higher incidence of diabetes and CHD in urban compared with rural India may reflect such a mechanism. The important issues of maternal insulin resistance during pregnancy, fetal growth, and future risk of diabetes and CHD in both mother and child need to be further investigated in prospective multiethnic studies. What constitutes improved nutrition for women before and during pregnancy also needs to be more clearly understood.

This study was funded by the Wellcome Trust, London, and the Medical Research Council (U.K.). We acknowledge the support of Sneha-India, the Indian society for research into the fetal and early life origins of adult disease.

We are grateful to the children and their families who took part; to biochemists Drs. S.S. Naik and D.S. Bhat; medical social workers M. Hoge and S. Bhangre; and staff nurses Ms. Gaikwad and Ms. Gokhale, who made significant contributions to the study. We thank Professor C.N. Hales and his staff in the Department of Clinical Biochemistry, Addenbrooke’s Hospital, Cambridge, U.K., for insulin assays and advice. We also thank Dr. T.D.R. Hockaday, Oxford, U.K., for comments on the draft.