Imprinted genes are epigenetically labile to early environmental influences 
. These mitotically heritable marks may serve as archives of early exposure and may offer a fascinating mechanistic explanation to the concept of developmental origins of adult disease. The best characterized imprinted domain is IGF2
that plays a key role in fetal growth, and alterations of methylation are present in fetal growth restriction 
and in Beckwidth-Wiedemann and Silver-Russell syndromes 
. In this study, we investigated two DMRs at the IGF2
locus in young adults born preterm at VLBW, who have previously been shown to have elevated levels of risk factors for CVDs in adulthood 
. We found that VLBW subjects have a significantly lower methylation level at one IGF2AS
DMR CpG than their term-born peers. The finding survived Bonferroni correction of multiple comparisons and was not explained by potential confounding factors.
Although VLBW birth and in utero
severe caloric restriction are not exactly similar early adverse events, our finding of 2% lower methylation in VLBW adults parallels that in the Dutch famine study, which showed that people who had been exposed to very early nutritional deprivation and had later increased incidence of chronic diseases had a 5% lower IGF2
methylation level – measured decades after the exposure – at the same CpG sites that we measured 
. Another evidence for epigenetic mechanisms influencing late-onset conditions comes from another study in which hypermethylation at some imprinted gene loci sensitive to prenatal nutrition were associated with myocardial infarction 
. Recent studies have shown associations between gestational age at birth and methylation at patterns in newborns at several genes 
. We are not aware of other reports on DNA methylation patterns in subjects born preterm at VLBW. Although one study has reported severe hypomethylation at IGF2/H19
locus in a term patient with birth weight less than -3SDS 
, another larger study on subjects born SGA did not reveal an association between IGF2
methylation levels and SGA birth 
. Accordingly, we did not find any difference in IGF2
methylation between those VLBW subjects that were born SGA and those born AGA. Etiologies of SGA births are multiple and there are potentially many different epigenetically regulated loci 
, as well as non-epigenetic factors 
that lead to later health outcomes in these subjects. Other perinatal characteristics, such as infection, bronchopulmonary dysplasia, or preeclampsia, were unrelated to methylation levels. However, we suppose our subject population is not large enough to permit us to properly assess these effects. In addition, it has only limited power to assess associations with adult phenotypic characteristics.
It is not possible to assure here that the altered methylation in IGF2
is not a primary defect that is responsible for the premature birth at VLBW. If the methylation change is secondary to VLBW birth, we can not be sure either that the 2% lower IGF2AS
CpG3 methylation can have any clinical relevance. Whether such a minor reduction in methylation can have functional consequence, such as downregulation of IGF2
function, is not known. However, methylation differences may be more significant in other tissues than the white blood cells from whole blood samples that we studied here 
. Analysis of methylation in, e.g., fat and muscle would give more detailed data for methylation associated with VLBW birth, but other tissues than blood were not available from our subjects. In addition to different tissues, methylation levels also vary across different blood cell types 
. Thus, the methylation levels studied could be affected by differential cell count in whole blood samples. Using a publicly available database 
, we checked the methylation levels for available CpGs within 1 kb from our amplicons, and did not find major methylation differences between blood cell types that would have affected the interpretation of our results.
A clear advantage of this study is that the study population was exceptionally well characterized allowing for, e.g., adjustment for most important potential confounders that could affect IGF2
methylation levels. It has been stated that infants born to smokers have higher methylation at the IGF2
than those born to non-smokers 
. Moreover, mother’s obesity before pregnancy may influence IGF2
methylation and gene expression 
. Although we observed a significant effect of mother’s smoking during pregnancy on methylation at some CpG units neither this nor any other confounding factor explained the differences in the IGF2AS
CpG3 methylation fraction between VLBW and control subjects. We did not adjust for mode of delivery, since the reasons for cesarean section are different in preterm and term pregnancies. In our study, mode of delivery was not associated with methylation levels. In vitro fertilization has been reported to affect methylation patterns 
, but it was not used in the 1970s. Ethnicity could also have an influence on methylation patterns, but our study participants were almost exclusively of Finnish origin. Folic acid supplementation that has been suggested to increase the risk of adult chronic diseases has been reported to be associated with methylation levels 
. We adjusted for folic acid use, but it did not change the results.
A minor disadvantage in our study is the number of study participants which was relatively small. Moreover, our study participants may not be representative of the original VLBW cohort, but this would only be expected to introduce bias if the association between VLBW birth and IGF2 methylation was different among participants and non-participants. This cannot be excluded.