BackgroundIntake of sweet and fatty snacks may partly contribute to the occurrence of obesity and other health conditions in childhood. Traditional dietary assessment methods may be limited in accurately assessing the intake of sweet and fatty snacks in children. Metabolite biomarkers may aid the objective assessment of children’s food intake and support establishing diet-disease relationships. ObjectiveThe present study aimed to identify biomarkers of sweet and fatty snack intake in two independent cohorts of European children. MethodsWe used data from the IDEFICS/I.Family cohort from baseline (2007/2008) and two follow-up examination waves (2009/2010 and 2013/2014). In total, n=1788 urine samples from 599 children were analysed for untargeted metabolomics using high-resolution liquid chromatography-mass spectrometry. Short-term dietary intake was assessed by 24-hour dietary recalls, and habitual dietary intake was calculated with the National Cancer Institute method. Data from the DONALD cohort of 24-hour urine samples (n=567) and 3-day weighted dietary records were used for external replication of results. Multivariate modelling with Unbiased Variable selection in R (MUVR) algorithms and linear mixed models were used to identify novel biomarkers. Metabolite features significantly associated with dietary intake were then annotated. ResultsIn total, 66 metabolites were discovered and found to be statistically significant for “chocolate candy”, “cakes, puddings & cookies”, “candy & sweets”, “ice cream”, and “crisps”. Most of the features (n=62) could not be annotated. Short-term and habitual chocolate intake were positively associated with theobromine, xanthosine, and cyclo(L-prolyl-L-valyl). These results were replicated in the DONALD cohort. Short-term “candy & sweets” intake was negatively associated with octenoylcarnitine. ConclusionWe identified potential metabolite biomarkers of sweet and fatty snacks in children, of which three biomarkers of chocolate intake, namely theobromine, xanthosine, and cyclo(L-prolyl-L-valyl) were externally replicated. However, these potential biomarkers require further validation in children.