A scheme of the number of participants and the investigations done at different time points is illustrated in Fig. 1.
During 1997–1998, 830 women booking consecutively into the antenatal clinic of the Holdsworth Memorial Hospital (HMH) in Mysore, south India, and matching our eligibility criteria (no known history of diabetes, intention to deliver at HMH, singleton pregnancy) had a 100 g 3 h OGTT at 30 ± 2 weeks’ gestation after an overnight fast . Weight, height and skinfold thickness (biceps, triceps, subscapular and suprailiac) were measured using standardised methods . Socioeconomic status (SES) was assessed using the Kuppuswamy score, a questionnaire which derives a score based on the education, occupation and the income of the head of the family ; a higher score indicates higher SES. The majority of our women belonged to middle or lower-middle social classes (score: 5–25).
Plasma glucose was measured using a standard hexokinase method, and insulin using a one-step chemiluminescent immunoenzymatic assay (Sanofi Pasteur Diagnostics, Marnes la Coquette, France). Complete OGTT results were available for 785 women. GDM was diagnosed in 49 women (6.2%) using the Carpenter–Coustan criteria , the standard method in clinical use in the HMH. Insulin resistance was estimated using the updated homeostasis model assessment equation (HOMA-2) from an online HOMA calculator . The women’s own consultant obstetricians managed their further clinical care.
A total of 630 women delivered live, normal babies in HMH. There were 41 women diagnosed with GDM, ten of whom received insulin treatment during pregnancy, but none of whom was on insulin or oral hypoglycaemic agents at the time of discharge from the hospital.
The hospital ethical committee approved the study, and informed verbal consent was obtained from all women.
Vitamin B12 and folic acid supplements
It was routine for general practitioners and obstetricians to prescribe folic acid and/or multivitamin supplements to pregnant women. Supplements taken by the women were recorded at recruitment but not subsequently at 30 weeks’ gestation, when blood samples were taken, or at term.
Vitamin B12 and folate during pregnancy
We used stored fasting plasma samples to measure vitamin B12 and folate in 774 of the 785 women who had complete OGTT data in pregnancy. The samples had been stored in the freezer (−80°C for 8 years) within 1 h of sampling, and were transferred on dry ice for laboratory analysis at the Diabetes Research Centre, KEM Hospital, Pune, India, using microbiological assays [16, 17]. Intra- and interassay coefficients of variations were <8% for both assays. As there is no specified cut-off for pregnancy, we defined vitamin B12 deficiency as a concentration <150 pmol/l and folate deficiency as a concentration <7 nmol/l based on the values generally used in a normal population [9, 18].
Follow-up of the women was based on the follow-up of their offspring . Twenty-five children died between birth and 5 years, seven children were excluded after birth because of medical reasons, and 43 families either refused follow-up or moved away from Mysore. Five years later, we were able to follow 555 women, 29 of whom were excluded from the current study because of recent pregnancy (within the previous 6 months). The remaining 526 women had a 2 h, 75 g OGTT; 519 of these women, for whom vitamin B12 and folate concentrations were measured (35 met criteria for GDM), completed the study. Detailed anthropometry was performed and systolic (SBP) and diastolic blood pressures (DBP) were measured using an automated BP monitor (CRITIKON, DINAMAP model 8100, Tampa, FL, USA). Plasma glucose (glucose oxidase–peroxidase method), triacylglycerols (glycerol 3 phosphate oxidase-peroxidase method) and HDL-cholesterol (direct HDL-cholesterol method) were measured on an autoanalyser (Alcyon 300; Abbott Laboratories, Abbott Park, IL, USA), and insulin was measured using a time-resolved, fluoroimmunoassay (DELFIA) method (PerkinElmer Life and Analytical Sciences, Wallac Qy, Turku, Finland) at the Diabetic Research Centre, Pune, India.
Diabetes was defined as a fasting glucose concentration ≥7.0 and/or 120 min glucose ≥11.1 mmol/l . Women were also classified as having diabetes if they had been diagnosed by a doctor as having diabetes since the index pregnancy. Impaired glucose tolerance (IGT) was a fasting glucose concentration <7.0 mmol/l and 120 min glucose ≥7.8 mmol/l but <11.1 mmol/l. Impaired fasting glucose (IFG) was defined as fasting glucose ≥6.1 mmol/l but <7.0 mmol/l .
Metabolic syndrome was defined by the International Diabetes Federation (IDF) criteria recommended for south Asian women . Waist circumference ≥80 cm, and any two of the following: triacylglycerol ≥1.7 mmol/l; HDL-cholesterol <1.29 mmol/l; SBP ≥130 or DBP ≥85 or having treatment for hypertension; fasting glucose ≥5.6 mmol/l; or type 2 diabetes.
The distributions of HOMA-2 and vitamin B12 concentrations were skewed; these data were log-transformed for analysis where required. The main exposures of interest were the vitamin B12 deficiency (yes/no), and plasma vitamin B12 and folate concentrations. Other confounding exposures such as maternal age, parity, religion, family history of diabetes and SES were used as covariates in the multiple regression models. The outcomes of interest were anthropometry, insulin resistance and the incidence of GDM during pregnancy, and anthropometry, insulin resistance and the prevalence of diabetes and metabolic syndrome at follow-up. Associations of maternal vitamin B12 and folate concentrations with anthropometry and HOMA-2 during pregnancy and at follow-up were examined using linear regression analysis, and with the incidence of GDM, and the prevalence of diabetes and metabolic syndrome at follow-up, using logistic regression analysis. Interaction terms were used to test for modification by folate status of associations between vitamin B12 status and the several diabetes-related outcomes by using vitamin B12 as two groups (deficiency and normal groups), and tertiles of folate concentrations. p values <0.05 were considered significant. All statistical analyses were performed using SPSS V16.