What children eat may leave a measurable signature in their blood

by · News-Medical

Researchers compared 236 blood metabolites with dietary patterns in European children, tracing how foods ranging from fish and olive oil to added sugars and saturated spreads corresponded with metabolic profiles.

Want to read later? Download your clean, ad-free, print-friendly PDF copy by clicking here.

A recent study published in the European Journal of Nutrition suggests that metabolomic profiling may help objectively characterize adherence to dietary patterns among children.

Background

Diet is key to a healthy lifestyle, but accurately assessing dietary intake remains difficult. Self-reported data is often unreliable. Among children, the uncertainty may be compounded by having to rely on family members and other caregivers for dietary information.

In addition to differences in dietary intake, it is also important to understand how metabolic differences among individuals affect the relationship between diet and health outcomes. Metabolic responses are influenced by genetics, age, biological differences in nutrient handling, the environment, and other lifestyle factors.

The current study used metabolomics to characterize blood metabolite profiles associated with three different dietary patterns in children.

Study characteristics

The study included 421 children from the EU Childhood Obesity Project with dietary and metabolomic data at 5.5 and/or 8 years of age; 156 had data at both ages. The three dietary patterns were identified from three-day food records at age 2, and the resulting pattern coefficients were applied at later time points to score adherence to these patterns across childhood.

These included Core foods (CORE), characterized by a high proportion of vegetables, fruits, fish, olive oil, and meat; animal protein sources (PROT), characterized by a high content of meat, fish, eggs, flavored milk, vegetables, and snacks; and poor-quality fats and sugars (F&S), characterized by high consumption of saturated spreads, added sugars, fruit juice, and soft cheese, with lower olive oil and fish intake.

Among 236 metabolites measured at both time points, 48 were associated with at least one dietary pattern: 23 with CORE, 18 with PROT, and 12 with F&S. The individual metabolite associations did not remain significant after false discovery rate correction, so the authors focused on those that remained consistent across metabolite ratios, metabolite groups, food groups, nutrients, and the two time points.

There were 23 CORE-associated metabolites, including 12 phosphatidylcholines (PCs) that often contained polyunsaturated fatty acids, especially docosahexaenoic acid (DHA), which is found mainly in oily fish. By comparison, higher ratios of docosapentaenoic acid (DPA) to DHA have been considered to reflect an inflammatory state and have been associated with obesity and other noncommunicable diseases (NCDs) in previous research.

Most PROT-associated metabolites were also PCs, especially long-chain polyunsaturated PCs. The PROT profile was negatively associated with shorter PCs and those containing saturated fatty acids.

The F&S pattern had the fewest associated metabolites and the greatest diversity.

Metabolite ratios and dietary profiles

The PC aa C40:5/PC aa C40:6 ratio, interpreted by the authors as reflecting the DPA/DHA ratio, was negatively associated with CORE, fish intake, and olive oil and positively associated with F&S. Its association with CORE was the only dietary-pattern ratio association that remained significant after false discovery rate correction. Both PROT and CORE were negatively associated with groups of medium- and long-chain PCs, while PCs enriched for saturated and monounsaturated fatty acids were positively associated with CORE.

The Fischer ratio (branched-chain AAs/aromatic AAs) was positively associated with CORE but negatively associated with F&S. It was also associated with healthy fats but negatively associated with sugar intake.

PROT was positively associated with several PCs, including some DHA-containing long-chain polyunsaturated PCs, and with some long- and very long-chain FFAs. By comparison, it was negatively associated with shorter or more saturated PCs. Both PROT and F&S were negatively associated with the long-chain/very-long-chain FFA ratio.

Before multiple-testing correction, all three dietary patterns were inversely associated with the ratio PC aa C38:3/PC aa C38:4.

Food groups and metabolites

The associations also corresponded in several cases to particular food groups and nutrients. Fish intake was positively associated with PUFA-rich PCs. LPC a C22:6 and the PC aa C40:5/PC aa C40:6 ratio were associated with healthy fats, defined in the study as fats from olive oil, nuts, and fish.

Several PCs were associated with total and animal protein, but negatively linked to carbohydrate intake. Grain intake (dominated by refined grains in this study) was associated with several LPCs and with a higher LPC/PC ratio. This ratio has previously been suggested as a marker of a pro-inflammatory state and is especially associated with refined grains.

Milk and yogurt were associated with PC aa C32:1 and PC ae C34:0, but negatively linked to the polyunsaturated/monounsaturated PC ratio. Dairy protein was associated with LPC e C16:0 and two sphingomyelins, but inversely with the polyunsaturated/saturated free fatty acid ratio.

Processed meat was inversely associated with metabolites related to the tricarboxylic acid cycle.

F&S had fewer individual metabolite associations and showed a generally opposite pattern to CORE in the metabolite-set analysis. It was positively associated with alanine and negatively associated with LPC a C22:6. It was also inversely associated with groups of long- and very-long-chain free fatty acids and odd-chain sphingomyelins.

Sensitivity analyses at 5.5 and 8 years showed generally consistent directions of association, though some appeared at only one age.

The authors reported that the three dietary patterns had distinct metabolomic signatures. Eight of the 23 metabolite groups were associated with one or more dietary patterns. In seven of the eight, the direction of association was reversed between CORE and F&S.

The contrasting association of LPC a C22:6 with CORE (positive) and F&S (negative) suggests that it might help distinguish higher-quality from poor-quality dietary patterns in children.

Limitations

The study involved multiple centers with detailed data collection and analysis at multiple levels, including adjustment for several confounders. The findings showed internal consistency across time points and analytic approaches. The results are biologically plausible and consistent with previous studies.

The study also has limitations. The high dimensionality of the metabolomics data relative to the cohort size was an important limitation. Most of the associations did not persist after being corrected for multiple testing. The identified associations should thus be treated as exploratory.

The targeted metabolomics approach and measurements performed in separate batches at the two ages might have excluded metabolites that could serve as dietary biomarkers, leaving some aspects of metabolism unexplored. The study also does not establish the causality of the observed associations.

Adherence scores at 5.5 and 8 years were based on dietary patterns derived at age 2, which may not fully capture age-related changes in dietary behavior. Some blood samples were obtained without fasting; all of these came from Germany, and adjustment for country may not remove all resulting bias.

Conclusions

These findings suggest that dietary patterns in children are associated with metabolite profiles, including individual metabolites and metabolite groups, at 5.5 and 8 years of age. According to the authors, this might indicate that metabolite profiles reflect habitual dietary intake.

Prominent signals included fish and healthy-fat intake, DHA-containing phosphatidylcholines, and the PC aa C40:5/PC aa C40:6 ratio, with broadly contrasting metabolomic patterns between CORE and F&S.

The convergence of findings across multiple analytic approaches supports their biological plausibility and reduces the likelihood that the signals reflect random statistical variation. Independent replication is required before these metabolites can be considered reliable biomarkers.

Journal reference: