In two separate contemporary birth cohorts we have identified associations between a common H19 2992 genotype and birthweight. This SNP is located in H19 exon 5 [see Additional file 2], however its effect on mRNA structure or function is unknown and H19 has no protein product. An insulator upstream of H19 [15–18] and enhancers downstream [13, 14] regulate imprinting of IGF2, which encodes a major fetal growth factor . In addition H19 RNA may suppress IGF2 expression in trans  and has tumour suppressor functions in cell transfection studies . Consistent with the birthweight association, H19 2992 genotype was also associated with cord blood IGF-II protein levels at birth.
Birthweight associations were found with both mother's and offspring's H19 2992 genotypes, and it is not clear whether our findings represent direct effects of mother's genotype, offspring genotype, or both. As in other population studies, the low number of available DNA samples and genotypes from fathers contributed to relatively lower statistical power to formally detect effects of H19 2992 allele transmission. However, we did observe significant birthweight and cord blood IGF-II associations with mother's untransmitted allele, suggesting direct effects of mother's genotype. While in mice H19 and IGF2 are not expressed postnatally , in humans IGF2 expression continues into adulthood . In most adult tissues IGF2 expression remains monoallelic as in the fetus, however biallelic expression is observed in adult human liver . It is possible that mother's H19 2992 genotype may regulate IGF2 expression in maternally-derived placental tissues. Alternatively, H19 regulation of maternal IGF2 expression could influence mother's glucose metabolism  and thereby influence placental glucose transfer and fetal growth . Offspring H19 2992 genotype was associated with birthweight, but not with IGF-II levels. It is possible that this discrepancy could relate to reduced power due to the smaller number of IGF-II samples available, or alternatively it could be possibly explained by a non-IGF-II mediated effect of H19 on birthweight. The association between birthweight and IGF-II levels in cord blood at birth is weak, and it is possible that the effects of IGF-II on fetal growth may be specific to certain tissues or developmental stages in fetal life .
The H19 2992 genotype associations with birthweight and IGF-II were independent of mother's body size and were more apparent in mother's first pregnancies. First-born offspring are more growth-restrained in utero than subsequent children, as evident by smaller birth size and compensatory rapid postnatal weight gain within the first 2 to 3 years . The mechanism of growth restraint seen in first pregnancies is unknown, but could reflect a maternally inherited genetic trait with subsequent relaxation of restraint in later pregnancies . We propose that the common H19 2992T allele, either in the mother or offspring, confers reduced fetal growth restraint particularly among first-born infants as indicated by larger birth size and less postnatal compensatory catch-up weight gain.
In subsequent pregnancies and in larger babies, a mendelian pattern of inheritance of birthweight is observed . It was in these second and subsequent pregnancies that we saw association between size at birth and common allele class at the insulin gene minisatellite (INS VNTR) , which also regulates IGF2 expression . Offspring INS VNTR class III alleles were associated with larger head size and higher IGF-II levels at birth, and this effect was masked in growth-restricted first pregnancies . Thus, there may be important interaction between maternal parity and imprinted and non-imprinted genes involved in the regulation of IGF-II expression and fetal growth.
Altogether we identified ten SNPs in H19. Given that there is no confirmed function for the H19 transcript in vivo , it is difficult to assess the effect of the H19 2992 SNP. It is possible that this variant: (a) may alter the RNA itself, and might influence the putative growth suppressing function of the H19 RNA; (b) may influence levels of H19 transcripts, leading to an effect as above; (c) may be in linkage with an unidentified regulatory element in H19 acting on IGF2 in cis; (d) may be in linkage with variants in known cis regulatory elements for IGF2, such as the insulator or the enhancers. Further work, including systematic genotyping of other SNPs in the extended region and functional studies, will be required to identify the causal variant(s) and, hence, possible mechanisms by which they may influence birthweight and cord blood IGF-II levels. The identification of separate effects of H19 on fetal growth, by acting both within the mother and also within the fetus, could be confirmed by studying a larger population to provide greater power to detect an interaction between mother's and offspring's genotypes, and by exploration of grand-parental genotype and allele transmission to the mother. Intriguingly, while we confirmed that the offspring 2992 genotype (CC vs. CT) association with birthweight was independent of mother's genotype (within CC mothers; Table 4), that finding also raises the surprising suggestion of a paternal H19 allele transmission effect, however, we did not detect any formal association with paternal transmission. An alternative explanation would be that the H19 2992 is in LD with another paternally transmitted determinant of birthweight. Further studies with larger numbers of complete informative trios will be needed to clarify this issue.