Another limitation of our study was the use of methylation profiles observed in blood. It is well known that tissue-specific DNA-methylation profiles exist; therefore, methylation profiles in blood might not be entirely representative of other tissues. If the primary interest for identifying epigenetic profiles is to determine causality, the tissue under investigation might be of great importance, and a more relevant tissue, such as adipose tissue, might be more suitable for a trait such as BMI or obesity. This might not apply for prediction, and comparing blood-derived methylation predictors with those derived from other tissues would be a logical next step if data were available.