Absence of Birth-Weight Lowering Effect of ADCY5 and Near CCNL, but Association of Impaired Glucose-Insulin Homeostasis with ADCY5 in Asian Indians

doi:10.1371/journal.pone.0021331

PLoS One. 2011; 6(6): e21331.

Published online 2011 June 21. doi: 10.1371/journal.pone.0021331

PMCID: PMC3119677

Copyright Vasan et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Absence of Birth-Weight Lowering Effect of ADCY5 and Near CCNL, but Association of Impaired Glucose-Insulin Homeostasis with ADCY5 in Asian Indians

Senthil K. Vasan,^1,²^* Matt J. Neville,³ Belavendra Antonisamy,⁴ Prasanna Samuel,⁴ Caroline H. Fall,⁵ Finney S. Geethanjali,⁶ Nihal Thomas,² Palany Raghupathy,⁷ Kerstin Brismar,¹ and Fredrik Karpe^2,^3,⁸^*

¹Department of Molecular Medicine & Surgery, Karolinska Institutet, Stockholm, Sweden

²Department of Endocrinology, Diabetes & Metabolism, Christian Medical College, Vellore, India

³Oxford Centre for Diabetes, Endocrinology and Metabolism, University of Oxford, Oxford, United Kingdom

⁴Department of Biostatistics, Christian Medical College, Vellore, Tamil Nadu, India

⁵Medical Research Council (MRC) Epidemiology Resource Centre, Southampton, United Kingdom

⁶Department of Clinical Biochemistry, Christian Medical College, Vellore, Tamil Nadu, India

⁷Department of Child Health, Christian Medical College, Vellore, Tami Nadu, India

⁸National Institute for Health Research (NIHR) Oxford Biomedical Research Centre, Oxford Radcliffe Hospitals (ORH) Trust, Oxford, United Kingdom

Murielle Bochud, Editor

University Institute of Social and Preventive Medicine, Switzerland

* E-mail: Fredrik.Karpe/at/ocdem.ox.ac.uk (FK); Email: vaskan/at/ki.se (SKV)

Conceived and designed the experiments: SKV BA CHF PR FK. Performed the experiments: MJN FK. Analyzed the data: SKV BA PS FK. Contributed reagents/materials/analysis tools: SKV MJN FK BA FSG NT KB. Wrote the paper: SKV FK.

Received January 12, 2011; Accepted May 26, 2011.

This article has been cited by other articles in PMC.

Abstract

Background

A feature of the Asian Indian phenotype is low birth weight with increased adult type 2 diabetes risk. Most populations show consistent associations between low birth weight and adult type 2 diabetes. Recently, two birth weight-lowering loci on chromosome 3 (near CCNL1 and ADCY5) were identified in a genome-wide association study, the latter of which is also a type 2 diabetes locus. We therefore tested the impact of these genetic variants on birth weight and adult glucose/insulin homeostasis in a large Indian birth cohort.

Methodology/Principal Findings

Adults (n

2,151) enrolled in a birth cohort (established 1969-73) were genotyped for rs900400 (near CCNL1) and rs9883204 (ADCY5). Associations were tested for birth weight, anthropometry from infancy to adulthood, and type 2 diabetes related glycemic traits. The average birth weight in this population was 2.79±0.47 kg and was not associated with genetic variation in CCNL1 (p

0.87) or ADCY5 (p

0.54). Allele frequencies for the ‘birth weight-lowering’ variants were similar compared with Western populations. There were no significant associations with growth or adult weight. However, the ‘birth weight-lowering’ variant of ADCY5 was associated with modest increase in fasting glucose (β 0.041, p

0.027), 2-hours glucose (β 0.127, p

0.019), and reduced insulinogenic index (β -0.106, p

0.050) and 2-hour insulin (β -0.058, p

0.010).

Conclusions

The low birth weight in Asian Indians is not even partly explained by genetic variants near CCNL1 and ADCY5 which implies that non-genetic factors may predominate. However, the ‘birth-weight-lowering’ variant of ADCY5 was associated with elevated glucose and decreased insulin response in early adulthood which argues for a common genetic cause of low birth weight and risk of type 2 diabetes.

Introduction

A recent genome-wide association (GWA) study identified two independent loci on chromosome 3 (rs900400 near CCNL1 and rs9883204 of ADCY5) for low birth weight [1]. When replicated in Europeans robust associations (p

3×10⁻²⁶ and 3×10⁻⁹, respectively) with birth weight were observed leading to a 113 g difference between homozygous carriers vs. non-carriers for the two respective variants [1].

Indian new born babies generally weigh the least in the world with a mean birth weight at term being 2.6 to 2.9 kg compared to 3.5 to 3.7 kg among Europeans [2], [3]. The discovery of genetic variants having a strong effect on birth weight therefore provides the opportunity to test if some of the low Indian birth weight has a genetic component.

A systematic review of the relationship between birth weight and subsequent risk of development of type 2 diabetes in adulthood has shown an inverse relationship that is consistent between populations [4]. Although the exact mechanism for this relationship remains unclear, the fetal-insulin hypothesis describes a distinct role for a genetic and environmental component for the association [5]. An alternative, but neither opposing, nor exclusive, explanation is through fetal programming due to the in utero environment [6]. However, this latter explanation remains mechanistically elusive.

Genetic variants that influence fetal insulin secretion or insulin sensitivity appear to be important determinants of fetal growth, and subsequent development of type 2 diabetes [7]. Associations between mutations in the INS, INSR, IPF1, KCNJ11, ABCC8 and HNF1B genes reducing fetal insulin secretion and markedly reducing birth weight in monogenic diabetes provide support for a genetic role in the modulation of birth weight, although this cannot be extrapolated to the general population due to the rarity of these mutations [5], [8]–[10]. There is some evidence from the study of common genetic variants related to type 2 diabetes (TCF7L, GCK, I/III polymorphisms of INS-VNTR and IGF-1) and its effects on birth weight, lending support to the notion that more moderate impairment of insulin secretion provides a link between inheritance of type 2 diabetes susceptibility and reduced weight at birth [11]–[14]. More recently genetic variants near ADCY5 have been linked to glycemic traits in healthy individuals [15], and with birth weight in the Danish Inter99 population [16].

Therefore, we aimed to replicate two ‘birth weight-lowering’ alleles (rs900400 near CCNL1 and rs9883204 of ADCY5) in a longitudinal birth cohort form South India and also test whether the low Indian birth weight is driven, or partly explained, by similar genetic factors as in Europeans. In this replication there are two inherent questions. First, whether the low birth in Indians is due to higher frequencies of ‘birth weight-lowering’ alleles compared to the Western population and secondly, whether the lower average birth weight is modulated by variants in ADCY5 and near CCNL1 in a way similar to the western population. Due to the known associations between birth weight and adult type 2 diabetes and the fact that one of the ‘birth weight-lowering’ alleles (ADCY5) also appears to be a genetic determinant of glycemic traits, we also aimed to investigate possible associations of CCNL1 and ADCY5 genotypes with diabetes related intermediate-traits in addition to investigating the influence of these SNPs on anthropometric traits during longitudinal follow-up from birth to adulthood.

Materials and Methods

Ethics Statement

All study participants gave written informed consent and the protocol was in accordance with the Helsinki Declaration, and approved by the Institutional Ethics committee and Review board of Christian Medical College & Hospital, Vellore, India.

Cohort description

The current study included 2,151 adults aged 26–32 (mean 28.3±1.10) years, drawn from an original birth cohort of 10,691 singleton births during 1969–1973 in one of 24 wards of Vellore town and adjoining rural villages, Tamil Nadu, India. Details of the original cohort are described elsewhere [17]. Complete birth measurements were available for 4,092 (2,790 from rural and 1,302 from Urban dwelling) individuals and 2,218 out of 2,572 members retraced during 1999–2002 agreed to take part in a study of glucose tolerance and cardiovascular risk factors, reported earlier [18]. The original cohort was followed up at several time points of growth and development from birth, through infancy (1–3 months), childhood (6–8 years) and adolescence (10–15 years) and adulthood. A total of 2,151 (1,175 of rural birth and 976 of urban birth) participants (out of 2,218 retraced) had adequate DNA samples and were included in the current study. Based on WHO criteria, there were 83 (3.9%) subjects with impaired fasting glucose (IFG) (>6.1 but below 7.0 mmol/l), 319 (14.8%) with impaired glucose tolerance (IGT) (2-hours blood glucose >8 mmol/l) and 55 (2.6%) with manifest type 2 diabetes (T2DM), demonstrated by an oral glucose tolerance test.

Anthropometric measurements and biochemical analysis

Anthropometric measurements included height, weight, waist and hip circumference measured by standard methods during adulthood. Body fat percentage was calculated from four skinfold measurements [19], [20]. Plasma glucose and insulin were recorded at 0, 30, 60 and 120 minutes, following a 75 g oral glucose tolerance test (after 12 hours of overnight fasting). Plasma glucose concentrations were measured by glucose oxidase/peroxidase methods, and serum lipids using commercial enzymatic kits (Roche Diagnostics, Germany) on a Hitachi 911 autoanalyser (USA). Plasma insulin was measured by an immunoradiometric assay using Coat-a-Count kits (Diagnostic Products Corporation, USA). The quality of these measurements was assessed using Roche Precinorm and Precipath controls for glucose and lipids and BioRad Lyphocheck Immunoassay controls for insulin. Intra- and inter-assay coefficients of variation for insulin estimations were 8.0–14.5 and 8.2–13.0% respectively. HOMA-IR and HOMA-B were calculated using an algorithm from http://www.dtu.ox.ac.uk/homacalculator/index.php. All biochemical measurements were done centrally at the Department of Clinical Biochemistry, Christian Medical College & Hospital, Vellore, on samples that were collected during adulthood (in 2002).

Genotyping

DNA was extracted from peripheral blood using Qiagen kits. The samples were genotyped using 10–20 ng genomic DNA in 384-well format on an ABI 7900 machine at final volume of 4 µl. The genotyping was performed using TaqMan® SNP Genotyping Assays C1860681_10 and C3035719_20 for rs900400 (near CCNL1) and rs9883204 (ADCY5), respectively. The TaqMan® genotyping master mix was used following the manufacturer's conditions. Genotyping quality control was tested in 8.6% of the samples (genotyped in duplicate) with 0% difference in genotype. Genotyping failed in 65 (3.0%) for rs900400 and 61 (2.8%) for rs9883204, most probably due to low quality DNA for the platform used.

Statistical methods

We tested the association between quantitative phenotypes and SNPs using ANOVA for normally distributed variables and Kruskal–Wallis test for skewed variables, adjusted for gender. Linear regression analysis was performed for each of the glycemic trait as the dependent variable against genotypes (additive model) as independent variables, with age, gender, consanguinity and BMI as covariates. Z-scores for selected anthropometric variables were age-adjusted and converted into within-cohort age- and sex-specific Z-scores [(subject mean- cohort mean)/cohort SD] considering variations in ages measured at different time points. The cohort mean and SD were derived from all 4,092 individuals with complete birth and parental measurements. Infant data were included if there was at least one measurement between 1 and 3 months, and the latest available time point used. Childhood data were included if there was at least one measurement between 6 and 8 years, and the average Z-score was used if there were more than one measurement. Adolescent data were included if there was at least one measurement between 10 and 14 years, the Z-score for the age closest to 12 years being selected. The main analysis used all available data at each time point. The absolute effect of genetic variants on Z-scores of anthropometry was assessed using ANOVA after adjustment for gestational age, gender and consanguinity. Effect size of the additive model was computed using regression models adjusted for gestational age, gender and consanguinity. All data were analyzed using STATA (Version 11.0). Subjects with T2DM (n

55) were excluded from all analysis. All diabetes related traits except 2 hours postprandial glucose and AUC glucose were log transformed to obtain normal distribution before analysis.

A power calculation based on variance explained by both the genetic variants (0.3% for CCNL1 and 0.1% for ADCY5) as previously described, showed a power of 74% and 32% to detect a variance (r²) explained, for CCNL1 and ADCY5 respectively using adjusted models, calculated at an alpha (0.05), and a sample size of 2,151. The coefficient of variation (CV) of birth weight was 18.8% for CCNL1 and 16.7% for ADCY5 and this was comparable to the effects demonstrated by these variants on birth weight as described by Freathy RM et al [1].

Results

There were no significant associations between either rs900400 (near CCNL1) or rs9883204 (ADCY5) with birth weight in the Indian population (Table 1). The allele frequencies of the ‘birth weight-lowering’ alleles were 0.21 for C allele (rs900400) and 0.81 for C allele (rs9883204). The rs9883204 showed some loss of expected heterozygosity and did not follow Hardy–Weinberg equilibrium (X2 4.45, p

0.035) in the cohort. This is a well-known phenomenon in cohorts with a high degree of consanguineous parentage, which was the case in 47% of the rural part of the cohort and 29% in the urban group.

Table 1

Demographic characteristics of the birth cohort.

Comparing birth weights in extreme homozygotes (CC) for both of the ‘birth weight-lowering’ variants, no statistically significant difference between carriers (2,772±535 g, n

62) and non-carriers (2,782±552 g, n

53, p

0.10) was observed.

We then compared the birth weights for participants with normoglycemia, IFG, IGT and manifest type 2 diabetes (T2DM) in the entire cohort. The mean birth weight among people with normoglycemia was 2,795±460 g (n

1,694), with IFG (n

83) 2,754±505 g, with IGT (n

319) was 2,763±496 g and 2,715±536 g in the diabetes group (n

55). The difference appeared largest between the normoglycemic group and the groups with manifest type 2 diabetes, but this was not statistically significant (p

0.24).

There was no significant relationship between either of the genotypes and Z-scores of anthropometry measured across various time periods at birth, infancy, childhood, adolescence or adulthood (Table S1). Neither of the SNPs in additive model showed significant association with any of the anthropometric measures of obesity in adulthood. (Table 1), and this signal was not picked up by either the BMI or waist circumference measurements. The additive model of ADCY5 variant was associated with raised 2-hour glucose postprandially (p

0.010) (Table 2).

Table 2

Association of glycemic traits with rs9883204 (ADCY5) additive model.

Regression analysis showed that the ‘birth weight-lowering’ variant of ADCY5, was significantly and positively associated with increase in 2-hour glucose, fasting glucose, reduced insulinogenic index and reduced 2-hour insulin (Table 2). Since the rise in 2-hour glucose is dependent on the rate of peripheral glucose disposal, we presumed that fasting glucose may confound this association. However, we found that the association of ADCY5 variant with 2-hour glucose remained significant even after adjusting for fasting glucose (Logged β 0.102, p

0.053). The ‘birth weight-lowering’ variants were not associated with calculated index of insulin resistance (HOMA-IR). No impairment of β-cell function was observed for variants at the CCNL1 locus (Table 3).

Table 3

Association of glycemic traits with rs900400 (CCNL1) additive model.

To further analyze the impact on postprandial glycemia in the homozygous carriers of the ADCY5 variant, we calculated the frequency of having plasma glucose above an arbitrary cut-off of 9 and 11 mmol/L, respectively. The lower level was chosen to enhance the capacity to detect even lower glycemic disturbances in these apparently healthy and young adults. At the 9 mmol/L cut off, the odds ratio (OR) and 95% confidence was 1.64 (1.08–2.48) p

0.016, whereas there was no significant effect at the 11 mmol/L cut off (OR 1.07, 95%CI 0.56–2.04, p=0.84).

Discussion

Genetic variants associated with a birth weight-lowering effect in other populations, near CCNL1 or in ADCY5 locus are not associated with birth weight in this Indian cohort, but the ‘birth weight-lowering’ variant of ADCY5 was associated with raised glucose and reduced early phase insulin secretion.

The Indian birth weight is generally about 0.8 kg less compared to the West [21], and the mean birth weight in our cohort was 2.8 kg. Not only was there any modulation between genotypes at this lower level, the gene frequencies of the ‘birth weight-lowering’ variants in ADCY5 and near CCNL were also the same (ADCY5) or lower (CCNL) than in the original report in Europeans [1]. The absence of an association between ‘birth weight-lowering’ genetic variants and birth weight is therefore likely to be attributed to strong environmental influences which dominate over the genetic effects. This would be supported by the observations of slightly higher birth weights in Indian immigrants in the West compared to the native Indian babies [22], [23]. The existence of other genetic variants influencing birth weight in the Indian setting is possible, but rather unlikely considering the considerable genetic diversity in India [24] contrasted with the uniformity in the low birth weight phenotype.

The ADCY5 variant has been associated with increased plasma fasting glucose at a genome-wide level [15] and it appears that, type 2 diabetes genes in general are reproduced in Indians [25]. In the current study, rs9883204 variant of ADCY5 was associated with raised fasting and 2-hour glucose concentrations following an oral glucose tolerance test performed in early adulthood. Although this finding would be in general agreement with the fetal-insulin hypothesis, it is paradoxical compared to a recent study in a large Danish cohort as another ‘birth weight-lowering’ (rs11708067) variant was associated with reduced adult insulin resistance [16]. The ADCY5 rs11708067 and rs9883204 are in close linkage disequilibrium in the Hapmap CEPH (Utah residents with ancestry from northern and western Europe) population data (D' 0.93, r2 0.72, Hapmap data rel 27 - http://hapmap.ncbi.nlm.nih.gov). rs9883204 is not represented in the Hapmap Indian population (GIH - Gujarathi Indian population), therefore LD cannot be calculated directly within Hapmap. However, a further SNP, rs17361324, which is a proxy for rs9883204 in the CEPH Hapmap data is also a proxy for rs11708067 in the GIH Hapmap data. This may suggest that, the high LD between these two SNPs in the CEPH data is also present in the Indian population and the difference does not appear to be explained by the genetic architecture and will need confirmation in future studies. Additionally, the low birth weight and T2DM risk allele of ADCY5 variant was associated with reduced early insulin response measured by insulinogenic index, consistent with original report by Freathy et al., supportive of a possible role in influencing insulin secretion [1]. It is proposed that ADCY5 risk allele may operate by different mechanisms by which they influence birth weight and T2DM risk susceptibility [16], the later probably may be through an effect on insulin secretion rather than insulin resistance. It is worth mentioning that for glucose and insulin, the association detailed in this cohort shows borderline significance and type 2 error cannot be excluded.

The loss of heterozygosity of rs9883204 in our population is probably related to the high endogamy which is commonly seen among Indians. It is generally assumed that an association detected by a well-powered GWAS is in LD with the functional variant. The strong signals in GWAS are related to the functional magnitude of the effect and therefore, we assume this holds good in our population despite a considerable degree of consanguinity.

Although, we have studied, the genetic variants associated with birth weight in one of the largest and homogenous birth cohorts from India, our study has limitations. Our study was adequately powered for CCNL1, but less strong for ADCY5 to explain the variance observed with these two SNPs. The CV for birth weight in our study was comparable to originally described and this further solidifies our power to detect an association. Also the lower mean birth weight, observed in this population, might potentially be associated with a reduced overall variance in the dependent variable, i.e., reduce the effect of the functional variant on birth weight, which could possibly contribute to the lowered power in our study. Although it is evident that a multiple comparison correction would abolish the modest associations with glycemic traits, we believe that the association is biologically relevant and that loss of statistical significance by multiple testing does not necessarily disprove a true association owing to the homogenous population studied within a small geographical region and the similar allele frequencies reported among Caucasians [1]. We did not have the maternal genotype to assess the effect of maternal genetic variants on birth weight due to non-availability of blood sample from the mothers. The low number of diabetes cases in this still young cohort did not allow for observing a possible link between diabetes incidence and birth weight or genetic links between them.

In conclusion, the ‘birth weight-lowering’ variants in ADCY5 and near CCNL1 showing strong associations with birth weight in European cohorts appear to have little or no effect in the Indian setting. However, the ‘birth weight-lowering’ variant in ADCY5 was associated with modest glucose intolerance in early adulthood which reinforces the argument for a genetic link between in utero growth and adult type 2 diabetes.

Supporting Information

Table S1

Association between rs900400 and rs9883204 and Z-scores of anthropometric traits from birth to adulthood.

(DOC)

Click here for additional data file.^{(81K, doc)}

Acknowledgments

The authors wish to thank Dr. Harvest Gu, Karolinska Institutet, Stockholm, Sweden for his support. We are grateful to all the study participants, health workers and the study team members.

Footnotes

Competing Interests: The authors have declared that no competing interests exist.

Funding: The study was supported by the British Heart Foundation. S.V. is on a scholarship from the Swedish Institutet, Stockholm, Sweden and also supported by Family Erling-Persson Foundation. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

1. Freathy RM, Mook-Kanamori DO, Sovio U, Prokopenko I, Timpson NJ, et al. Variants in ADCY5 and near CCNL1 are associated with fetal growth and birth weight. Nat Genet. 2010;42:430–5. [PMC free article] [PubMed]

2. Sachdev HPS. Low birth weight in south asia. In: Gillepspie S, editor. Malnutrition in South Asia: A regional profile. New York: NY: United Nations Children Fund; 1997. pp. 23–50.

3. Kinare AS, Chinchwadkar MC, Natekar AS, Coyaji KJ, Wills AK, et al. Patterns of fetal growth in a rural Indian cohort and comparison with a Western European population: data from the Pune maternal nutrition study. J Ultrasound Med. 2010;29:215–23. [PMC free article] [PubMed]

4. Whincup PH, Kaye SJ, Owen CG, Huxley R, Cook DG, et al. Birth weight and risk of type 2 diabetes: a systematic review. JAMA. 2008;300:2886–97. [PubMed]

5. Hattersley AT, Tooke JE. The fetal insulin hypothesis: an alternative explanation of the association of low birth weight with diabetes and vascular disease. Lancet. 1999;353:1789–92. [PubMed]

6. Barker DJ. The origins of the developmental origins theory. J Intern Med. 2007;261:412–7. [PubMed]

7. Freathy RM, Bennett AJ, Ring SM, Shields B, Groves CJ, et al. Type 2 diabetes risk alleles are associated with reduced size at birth. Diabetes. 2009;58:1428–33. [PMC free article] [PubMed]

8. Støy J, Edghill EL, Flanagan SE, Ye H, Paz VP, et al. Insulin gene mutations as a cause of permanent neonatal diabetes. Proc Natl Acad Sci U S A. 2007;104:15040–4. [PubMed]

9. Edghill EL, Bingham C, Slingerland AS, Minton JA, Noordam C, et al. Hepatocyte nuclear factor-1 beta mutations cause neonatal diabetes and intrauterine growth retardation: support for a critical role of HNF-1beta in human pancreatic development. Diabet Med. 2006;23:1301–6. [PubMed]

10. Babenko AP, Polak M, Cavé H, Busiah K, Czernichow P, et al. Activating mutations in the ABCC8 gene in neonatal diabetes mellitus. N Engl J Med. 2006;355:456–66. [PubMed]

11. Freathy RM, Weedon MN, Bennett A, Hypponen E, Relton CL, et al. Vol. 80. Am J Hum Genet; 2007. Type 2 diabetes TCF7L2 risk genotypes alter birth weight: a study of 24,053 individuals. pp. 1150–61. [PubMed]

12. Weedon MN, Frayling TM, Shields B, Knight B, Turner T, et al. Genetic regulation of birth weight and fasting glucose by a common polymorphism in the islet cell promoter of the glucokinase gene. Diabetes. 2005;54:576–81. [PubMed]

13. Vaessen N, Janssen JA, Heutink P, Hofman A, Lamberts SW, et al. Association between genetic variation in the gene for insulin-like growth factor-I and low birthweight. Lancet 23; 2002;359:1036–7. [PubMed]

14. Lindsay RS, Hanson RL, Wiedrich C, Knowler WC, Bennett PH, et al. Vol. 52. Diabetes; 2003. The insulin gene variable number tandem repeat class I/III polymorphism is in linkage disequilibrium with birth weight but not Type 2 diabetes in the Pima population. pp. 187–93. [PubMed]

15. Dupuis J, Langenberg C, Prokopenko I, Saxena R, Soranzo N, et al. New genetic loci implicated in fasting glucose homeostasis and their impact on type 2 diabetes risk. Nat Genet. 2010;42:105–16. [PMC free article] [PubMed]

16. Andersson EA, Pilgaard K, Pisinger C, Harder MN, Grarup N, et al. Type 2 diabetes risk alleles near ADCY5, CDKAL1 and HHEX-IDE are associated with reduced birthweight. Diabetologia. 2010;53:1908–16. [PubMed]

17. Antonisamy B, Raghupathy P, Christopher S, Richard J, Rao PS, et al. Vol. 38. Int J Epidemiol.; 2009. Cohort Profile: the 1969-73 Vellore birth cohort study in South India. pp. 663–69. [PMC free article] [PubMed]

18. Raghupathy P, Antonisamy B, Geethanjali FS, Saperia J, Leary SD, et al. Glucose tolerance, insulin resistance and insulin secretion in young south Indian adults; relationships to parental size, neonatal size and childhood body mass index. Diabetes Research Clin Pract. 2010;87:283–92. [PMC free article] [PubMed]

19. Durnin JVGA, Womersley Body fat assessed from total body density and its estimation from skin fold thickness; measurements on 481 men and women aged from 16 to 72 years. Br J Nutr. 1974;32:77–97. [PubMed]

20. Kuriyan R, Petracchi C, Ferro-Luzzi A, Shetty PS, Kurpad AV. Validation of expedient methods for measuring body composition in Indian adults. Indian J Med Res. 1998;107:37–45. [PubMed]

21. Leary S, Fall C, Osmond C, Lovel H, Campbell D, et al. Geographical variation in neonatal phenotype. Acta Obstet Gynecol Scand. 2006;85:1080–9. [PMC free article] [PubMed]

22. Harding S, Rosato MG, Cruickshank JK. Lack of change in birthweights of infants by generational status among Indian, Pakistani, Bangladeshi, Black Caribbean, and Black African mothers in a British cohort study. Int J Epidemiol. 2004;33:1279–85. [PubMed]

23. Dhawan S. Birth weights of infants of first generation Asian women in Britain compared with second generation Asian women. BMJ. 1995;311:86–8. [PMC free article] [PubMed]

24. Reich D, Thangaraj K, Patterson N, Price AL, Singh L. Reconstructing Indian population history. Nature. 2009;461:489–94. [PMC free article] [PubMed]

25. Chauhan G, Spurgeon CJ, Tabassum R, Bhaskar S, Kulkarni SR, et al. Impact of common variants of PPARG, KCNJ11, TCF7L2, SLC30A8, HHEX, CDKN2A, IGF2BP2, and CDKAL1 on the risk of type 2 diabetes in 5,164 Indians. Diabetes. 2010;59(8):2068–2074. [PMC free article] [PubMed]

Articles from PLoS ONE are provided here courtesy of
Public Library of Science