Several GWAS and large-scale meta-analysis for adult BMI have identified susceptibility loci for increased BMI, explaining only a small proportion of variation in BMI 
. Investigation of genetic risk factors for childhood obesity has been restricted mainly to the replication of GWAS findings 
. Though modern genotyping arrays capture most of the common variations, there remain substantial additional contributions to phenotypic variation from the variants not well captured by these arrays 
. Investigation of variants prioritized based on their functional and positional significance provides an opportunity to identify those not detected by GWAS. Employing functional candidate analysis, we recently identified association of common variants of IL6
, and PBEF1
with increased susceptibility to obesity and measures of adiposity in Indian children 
. Homocysteine metabolism pathway seems to be a promising candidate pathway by virtue of its involvement in important cellular processes. Thus, here we comprehensively examined association of common variants of homocysteine metabolism pathway genes with obesity in children.
The study reveals association of AMD1
variant rs2796749 with obesity, measures of adiposity and plasma leptin levels in urban Indian children. To the best of our knowledge, this is the first study demonstrating the association of AMD1
variant with obesity. In our previous study of association of homocysteine metabolism pathway genes with type 2 diabetes and adult obesity 
, rs2796749 did not show association with adult obesity or BMI but was directionally consistent. Further, we investigated effect of this variant on adult BMI using the association data from GIANT consortium GWA meta-analysis which included 249,796 individuals of European ancestry 
. In this recent meta-analysis also, rs2796749 showed consistency in direction of association with BMI, though could not reach statistical significance (P
0.19). This suggests that AMD1
variant rs2796749 might have influence on adult BMI also, but with smaller effect size.
codes for an important regulatory enzyme adenosyl-methionine decarboxylase 1 (AdoMetDC) which catalyzes the decarboxylation of S-adenosyl methionine (SAM), a methyl-group donor in transmethylation reactions. The imbalances in decarboxylation of SAM might result in disruption of DNA methylation process leading to abnormal gene expressions 
, which is implicated in various metabolic disorders including obesity and diabetes 
. In addition to its role in transmethylation reaction, AdoMetDC is also a regulatory enzyme in polyamine biosynthesis. SAM is committed to polyamine biosynthesis after decarboxylation and serves as an aminopropyl donar in spermidine and spermine synthesis 
. The polyamines are ubiquitous and are implicated in cellular growth and differentiation, regulation of adipocyte formation and glucose metabolism 
A recent study reported increased levels of polyamines in childhood obesity and their correlation with biomarkers of oxidative stress, inflammation and leptin 
. The study supported the role of polyamines on growth and development on adipose tissue. On similar lines, our study demonstrated association of rs2796749 in 5? flank region of AMD1
with obesity in Indian children. The same variant also showed association with quantitative measures of adiposity including weight, BMI, hip and waist circumferences and plasma levels of leptin. Moreover, AMD1
mRNA is shown to increase in response to insulin via insulin response element in 5' flank region of AMD1
. The variant rs2796749 might also have role in influencing the transcription of AMD1
in response to insulin, which needs to be investigated.
codes for methionine adenosyltransferase (MAT), which catalyzes the conversion of methionine to S- adenosylmethionine (SAM). MAT1A
variants have been shown to be associated with hypertension and stroke 
. Our study also suggested association of MAT1A
variant rs1985908 with childhood obesity. MAT1A
variants also showed association with the plasma levels of leptin, adiponectin and triglyerides. But these associations could not stand multiple testing corrections, hence further replication in larger sample set and in other ethnic groups are required to substantiate the observations in the present study.
Although the present study design limits the potential to identify the causal relationships, based on previous studies, we can speculate that AMD1
variant might influence the susceptibility to obesity by modulating either the polyamines metabolism or DNA methylation. Dietary habits can affect these processes as nutrients such as vitamin B12 and folate are required as important cofactors in these enzymatic reactions (). The plasma levels of folate and vitamin B12 as well as the dietary intake of these are inversely correlated with the plasma homocysteine levels 
. Moreover, individuals consuming only vegetarian diet tend to have low vitamin B12 and elevated homocysteine levels 
. Majority of Indian population adheres to a vegetarian diet that predisposes them to vitamin B12 deficiency. Thus diet which can influence the availability of vitamins may modulate the impact of genetic variants on obesity phenotypes. Previous studies including investigation in Indian populations suggested the possibility of interaction between diet and genetic variants of homocysteine pathway genes in influencing the homocysteine levels 
, however no study provided strong evidence for this interaction and its impact on impact on phenotypes. Unfortunately, lack of data on the levels of folate and vitamin B12 and the dietary habits of our study subjects limited further investigation of the interaction of these factors with the genetic variants and their combined effect on obesity phenotypes.
In case-control association studies, limited statistical power and population stratification can lead to spurious association results. Though the sample size of our study is considerable, there is a likelihood of false negative observations for variants with small effect sizes, as the present study is sufficiently powered to capture only large effects (OR>1.3) of the common variants with frequencies more than 0.20. Further, to minimize the effect of population stratification, we have collected samples from a small geographical region that forms a homogenous cluster as reported by the Indian Genome Variation Consortium 
. Moreover, the multidimensional scaling (MDS) analysis using the genotype data for 595 unlinked markers (r2
<0.20) for stage 1 samples shows that the study population belongs to one cluster (Figure S1
In conclusion, we demonstrate here, for the first time, association of AMD1 variant with susceptibility to obesity and measures of adiposity in Indian children. Further studies to confirm the association of AMD1 variant with childhood, its functional significance and molecular mechanism need to be undertaken. Moreover, our study provides a lead for future investigations towards understanding the genetic predisposition of obesity in childhood and exploring therapeutic options for prevention of childhood obesity.