Vitamin B₁₂ and folate concentrations during pregnancy and insulin resistance in the offspring: the Pune Maternal Nutrition Study

The recruitment of mothers and follow-up of the children in the PMNS (Fig. 1) has been described [6, 8, 9]. The study started in 1993. Non-pregnant married women living in six villages near Pune had detailed anthropometric measurements every 3 months and menstrual dates were recorded monthly. Women who became pregnant were assessed at 18 ± 2 and 28 ± 2 weeks of gestation for anthropometry, physical workload, dietary intake (24 h recall and food frequency questionnaire [FFQ]) and measurement of circulating nutrients. According to the National Nutritional Anemia Control Programme, women were given 100 tablets of iron (60 mg per tablet) and folic acid (500 μg per tablet) from 18 weeks of gestation. Newborn babies were measured within 72 h of birth (1994–1996), and the children were followed up post-natally with repeat anthropometry every 6 months.

At 6 years of age (December 2000–February 2003) the children were investigated for body size, body composition and risk factors for type 2 diabetes mellitus and cardiovascular disease. Families were instructed to continue their usual diet and activity prior to the study, and were transported to the Research Unit the previous evening. They were provided with a standard dinner and then water only until the next morning. A fasting blood sample was collected, followed by a 1.75 g/kg body weight oral anhydrous glucose load and a further blood sample 120 min later.

The family’s socioeconomic status (SES) was assessed using the Standard of Living Index devised by the National Family Health Survey [13]. Weight and height were measured using standardised protocols. Whole-body dual-energy X-ray absorptiometry (DEXA) scans (Lunar DPX-IQ 240; Lunar Corporation, Madison, WI, USA; paediatric software) were carried out to measure total, truncal and leg fat mass, percentage body fat and lean mass. Standard quality assurance tests were performed every day. The DEXA machine was installed 3 months after the study started, and therefore 59 children had DEXA measurements later than the other investigations.

Plasma glucose was measured using an Hitachi 911 automated analyser (Hitachi, Tokyo, Japan) by the glucose oxidase peroxidase method. Plasma insulin was measured using a Delfia technique (Victor 2; Wallac, Turku, Finland) [14]. Insulin resistance was calculated using the fasting insulin and glucose concentrations (homeostatic model assessment of insulin resistance [HOMA-R]) [15]. As previously described, maternal erythrocyte folate concentrations were measured at the time of the original study (1993–1996), taking all necessary precautions in the collection, transport and storage of samples [8]. Based on the findings of a preliminary study [10], we measured plasma vitamin B₁₂, tHcy and methylmalonic acid (MMA) concentrations in all the stored maternal fasting samples (−80°C) in February 2004, as described [11].

Ethical permission for the study was granted by the KEM Hospital Ethical Committee, and by the local village leaders. Parents gave informed written consent. None of these volunteers were paid for participating in the study.

As a final stage of the analysis, we summarised the inter-relationships between maternal factors and outcomes in the children using principal components analysis (PCA) and conditional independence analysis. We first performed a PCA to condense the information contained in large groups of variables into a small number of ‘component’ variables. These components are independent of each other and therefore can be used in regression analysis without the problem of colinearity [18]. The groups of variables were: (1) maternal pre-pregnant size (height, skinfold thicknesses, head circumference, mid-upper-arm circumference [MUAC] and waist and hip circumferences); (2) maternal macronutrient intakes (energy, protein, carbohydrate and fat intakes at 18 and 28 weeks); (3) maternal micronutrient-rich foods in pregnancy (frequency of intake of dairy products, GLV, fruit and non-vegetarian items [meat, fish and eggs] at 18 and 28 weeks of gestation); (4) maternal micronutrient status (vitamin B₁₂, folate, tHcy and MMA concentrations at 18 and 28 weeks); (5) newborn size (birthweight, length, skinfold thicknesses, MUAC, head and abdominal circumferences); and (6) 6 year body composition (height and DEXA measurements of lean mass, and total, truncal and limb fat). Other key variables (maternal SES and physical workload score [mean of 18 and 28 week scores], and HOMA-R in the child at 6 years) remained as single variables. These, and the relevant principal components were included simultaneously in a conditional independence analysis, which is a method of displaying ‘pathways’ of association between a pre-specified set of variables (in this case, the groups of variables described above). The partial correlation (that is, the correlation while holding the remaining variables constant in the set) of each pair of variables was calculated [19]. A ‘path’ diagram was then drawn, connecting pairs of variables that were significantly correlated (p < 0.01). Two-tailed significance was calculated at 5% level. Analyses were performed using SPSS 11.0 for windows (SPSS, Chicago, IL, USA) and STATA 7.0 (STATA, College Station, TX, USA).