Infants and parents were examined by the medical officer and questioned regarding any major illness in the past and recent medical problems. Blood pressure was measured with a standard sphygmomanometer by the medical officer. The mean of three readings of systolic and diastolic pressure was recorded. Hypertension was defined as systolic pressure above 130 mm Hg, and diastolic pressure above 85 mmHg. After an overnight fast, oral Glucose tolerance test was done. A fasting blood sample was collected, and 30 and 120 minute samples were collected after glucose administration. Lipid profile, consisting of total cholesterol, LDL, VLDL, HDL and triglycerides, was measured (enzymatic method). Insulin was measured by Delfia technique and ELISA. Homeostatic Model Assessment – Insulin Resistance (Homa-IR) was calculated using the online Oxford model (http://www.dtu.ox.ac.uk).

Weight was measured by an electronic scale with an accuracy of ±10g. Standing and sitting height was measured to the nearest 0.5 cm by a wall-mounted stadiometer using the standard technique, described by Tanner [8]. BMI was calculated and categorized. Waist circumference was measured by a non-stretchable tape to the nearest 0.1 cm midway between the lower costal margin and superior iliac crest in expiration. Hip circumference was measured at the point of maximum protuberance. Skin fold thickness was measured at 4 sites – biceps, triceps, subscapular and suprailiac, using Harpender’s caliper. Dietary intake was assessed by 24- hour diet recall (weekday and weekend) and validated food frequency questionnaire [9]. Physical activity was assessed by using a questionnaire and BMR; physical activity level (PAL) was calculated using physical activity ratio (PAR) values of each activity in 24 hours [10].

Statistical analysis: For the variables not normally distributed, appropriate transformations for the underlying normality assumption are used. The linear association between the normally distributed variables was assessed by Pearson’s correlation coefficients; otherwise Spearman’s correlation coefficients were used. The partial correlation analysis was also used to test the independent associations between several variables of interest. The factors such as sex, socio-economic status and current BMI were used as confounders for most of the correlation analysis.

The LBW cases and the control groups were first compared by using independent sample ‘t’ test for quantitative variables. Simple Chi-square test or Fisher’s exact test for independence of attributes was used to explore the differences between the groups in the prevalence of specific components of metabolic syndrome.

We have previously described the growth and cognitive development of 161 LBW infants at 18 years [5,6]. There was a dropout of 8 LBW subjects at 22 years, and hence 153 LBW and 77 normal controls were studied. The birth weight of the study group ranged from 866 g to 1999 g with a mean (SD) of 1545.5 (243.9) g and the mean (SD) birth weight of the control group was 2835.0 (305.8) g. There were 93 (60.8%) males and 60 (39.2%) females in the LBW group and 45 (58.4%) males and 32 (41.1%) females in the control group. There were 60 (39.2%) preterm SGA, 38 (24.8%) full term SGA and 55 (35.9%) preterm AGA babies in the LBW cohort. The comparison of the anthropometric measurements and biochemical parameters and MRI fat between cases and controls is shown in Table I. There was significant difference between the blood pressure of cases and controls.

TABLE I	Comparison of Anthropometry, Biochemical, Clinical and Imaging Parameters Between Cases and Controls (N=230)

Higher BMI at 1, 2, 6, 12, 18 and 22 years correlated significantly with higher systolic pressure. Higher BMI at 12 years correlated significantly with higher fasting insulin and HOMA-IR at 22 years. Physical activity and dietary intake was correlated with the above mentioned parameters, after adjusting for sex and current BMI. The 120-minute insulin level showed inverse correlation with physical activity. Abdominal obesity determined by MRI estimation was not significantly different between cases and controls (Table I).

Table II presents the results of multiple linear regression analysis with 120-minute glucose as the dependent variable. Sum of four skinfold thickness at 22 years was the significant and independent determinant of 120-minute glucose. A similar regression analysis for independent determinants of Homa-IR at 22 years also showed that sum of four skinfold thickness was a significant determinant (Table II). Similar results were seen for triglycerides. The 22-year BMI was a significant independent determinant for blood pressure.

TABLE II  Independent Determinants of 120-minutes Glucose  and HOMA-R at 22 years 

The birthweight and 22-year weight was divided into four quartiles. Those who were born small and became big at 22 years were compared with those who were born small and remained small (Table III) at 22 years. Fasting, 30 min, 120 min insulin and HOMA-R were significantly higher in those born small and became big at 22 years. Serum triglycerides and cholesterol were also higher in this group. They had higher systolic blood pressure and had more abdominal obesity on MRI (Table III).

TABLE III Parameters of  ‘Small at Birth and Big at 22 Years’ and ‘Small at Birth and Small at 22 Years’ Subjects

There were only six cases of polycystic ovarian syndrome. None of the females had hirsutism. The preterm AGA and SGA cases were compared for all biochemical parameters and blood pressure and MRI fat (Web Table I). There was no difference in any of the parameters, except for systolic blood pressure.

The BMI of both mother and father was compared separately with that of the cases and controls. The BMI of parents correlated with BMI of the study group. When the incidence of the five components of Met-S between those having family history of diabetes and cardiovascular disease and those not having a family history was compared, there was no significant difference. During our study, six parents were found to be prediabetic, two were frank diabetic and five parents had hypertension, and were not aware of it.

This study showed that the prevalence of three components of metabolic syndrome was low at 22 years in LBW children. However, when only two components were concerned, there was a significant difference between their occurrence in cases and controls, hypertension being the first component to appear. The study also showed that ‘small at birth and big at 22 years’ had many adverse effects like hypertension, increased cholesterol and triglycerides, increased insulin levels and insulin resistance, and increased abdominal obesity. Sum of skinfold thickness at 22 years was an independent determinant of 120-minute glucose, insulin resistance and triglycerides.

This is the last phase of a long prospective study spanning 22 years. In order to enroll the 22-year-old subjects, we offered a free check-up for parents as an added incentive. The dietician could assess the diet of the family and give advice regarding appropriate diet. The main limitation of this study was the less number of controls compared to the cases.

We had earlier studied these subjects for central obesity by taking all their anthropometric measurements for adiposity at 18 years, and found no evidence of adiposity [5]. There are several reports that Asians have normal BMI, but have excess deposition of fat around the waist [14]. The Rancho Bernardo study [16] showed that low birth weight coupled with adult obesity is a strong determinant of metabolic syndrome in postmenopausal women. We found that BMI at 12, 18 and 22 years significantly correlated with higher insulin and HOMA-IR. Sachdev, et al. [16] reported that serial measurements of childhood BMI give useful prediction regarding risk of developing adultmetabolic syndrome. Higher BMI at 1, 2, 6 and 18 and 22 years correlated with higher systolic blood pressure in our study. A lot of stress has been placed on BMI in many of the studies on the metabolic syndrome, but very little attention has been given to measurements of adiposity. Dalleck, et al. [17] found that waist circumference was independently associated with HDL cholesterol. We found sum of four skinfold thicknesses to be relatively a stronger correlate for biochemical parameters as well as blood pressure compared to BMI, waist circumference and MRI estimation of subcutaneous abdominal fat. A simple measurement of adiposity such as sum of four skinfold thickness may be sufficient, compared to a time consuming and expensive measurement like MRI. Lloyd, et al. [18] carried out a meta-analysis of all studies that linked childhood obesity with risk of developing metabolic syndrome. They also reported that there is a little evidence to suggest that greater BMI in childhood was an independent factor for dyslipidemia in adulthood.

We found that "small at birth and big at 22 years" had higher insulin levels and high HOMA-IR. A study from Denmark [19] showed that weight gain in first 3 months of life may increase the risk of metabolic syndrome, particularly glucose metabolism in SGA children. A large study from China [20] reported several components of metabolic syndrome presenting in LBW children in late adulthood. Bavdekar, et al. [21] reported "small at birth and big at 8 years" to have impaired glucose tolerance test. Although we had very few cases with three components to meet the IDF criteria of Metabolic Syndrome, the LBW cases had a higher number of two components compared to controls at 22 years. Our findings suggestive of early onset of trends of meatbolic syndrome need further study and reassessment again after a few years. However, it definitely points out the need to initiate early interventions of dietary and lifestyle influences.

Contributors: SC: conceived and carried out the study and wrote the manuscript and shall be guarantor for the paper; MO: helped in data collection and counseling of the family; MH: was responsible for getting the patients and collecting blood samples; AP: supervised the project; MS: statistical analysis. All authors contributed to manuscript writing and its approval.

Funding: Indian Council of Medical Research, New Delhi, India. Grant No. 5/4/8-7/2009-NCD-II.

Competing interest: None stated.

1. Ramachandran A, Snehalatha C, Latha E, Vijay V, Viswanathan M. Rising prevalence of NIIDM in an urban population in India. Diabetologia. 1997;40:232-7.

2. Barker DJP. Foetal origins of coronary heart disease. BMJ. 1995;311:171-4.

3. Reddy KS. Cardiovascular diseases in India. World Health Stat. 1993;46:101-7.

4. Sachdev HPS. Nutritional transition in the backdrop of the early life origins of adult disease: a challenge for the future. Indian J Med Res. 2004;119:iii-v.

5. Chaudhari S, Otiv M, Khairnar B, Pandit A, Hoge M, Sayyad M. Pune low birth weight study – Growth from birth to adulthood. Indian Pediatr. 2012;49:727-32.

6. Chaudhari S, Otiv M, Khairnar B, Pandit N, Hoge M, Sayyad M. Pune low birth weight study-Birth to adulthood – cognitive development. Indian Pediatr. 2013;50:853-7.

7. Singh M. Care of the Newborn. 4th ed. New Delhi: Sagar Publications, 1979.

8. Tanner JM. Normal growth and techniques of growth assessment. Cl Endocr and Metab. 1986;3:411-51.

9. Gopalan C, Rama Shashtri BV, Balsbramanian SC. Nutritive Value of Indian Foods. National Institute of Nutrition, Indian Council of Medical Research, 2002.

10. Kurpad AV, Swaminathan S, Bhat S. IAP National Task Force for Childhood Prevention of Adult Diseases: the effect of childhood physical activity on prevention of adult diseases. Indian Pediatr. 2004;41:37-62.

11. Golan R, Shelef I, Rudich A, Gepner Y, Shemesh E, Chassidim Y, et. al. Abdominal superficial subcutaneous fat: A putative distinct protective far subdepot in Type 2 diabetes. Diabetes Care. 2012;35:640-7.

12. Rumack CM, Wilson SR, Charboneau JW. Diagnostic Ultrasound. Vol I, 2nd ed. Mosby 1998. P. 313.

13. The IDF Consensus Worldwide Definition of the Metabolic Syndrome Available from :https://www.idf.org/webdata/docs/IDF_Meta_def_Final.pdf.

14. Mishra A, Khurana L. Obesity and the metabolic syndrome in developing countries. J Clin Endocrinol Metab. 2008;93:509-25.

15. Yarbrough D, Barrott ConnorE, Krietz Silverstein, Wingard DL. Birth weight, adult weight, and girth as predictors of metabolic syndrome in postmenopausal women: the Rancho Bernardo Study. Diabetes Care.1998;21:1652-8.

16. Sachdev HP, Osmond C, Fall CH, Lakshmy R, Ramji S, Dey Biswas SK, et. al. Predicting adult metabolic syndrome from childhood body mass index: follow-up of the New Delhi cohort. Arch Med Res. 2008;39:796-802.

17. Dalleck LC, Kiellan EM. The prevalence of metabolic syndrome and metabolic syndrome risk factors in college-aged students. Am J Health Promot. 2012;27:37-42.

18. Lloyd LJ, Langley-Evans, McMuellen S. Childhood obesity a risk of adult metabolic syndrome: a systematic review. Intl J Obesity. 2012;36:1-11.

19. Fabricius-Bjerre S,　Jensen RB,　Færch K,　Larsen T,　Mølgaard C,　Michaelsen KF,　et al. Impact　of　birth weight　and　early　infant　weight　gain on insulin resistance and associated cardiovascular risk factors in adolescence. PLoS One.　2011;6:e20595.

20. Xiao X, Zhang ZX,　Li WH,　Feng K,　Sun Q,　Cohen HJ,　et al. Low birth weight is associated with components of the metabolic syndrome. Metabolism.　2010;59:1282-6.