Metabolomic heterogeneity of ageing with ethnic diversity: a step closer to healthy ageing.

INTRODUCTION: Outside of case-control settings, ethnicity specific changes in the human metabolome are understudied especially in community dwelling, ageing men. Characterising serum for age and ethnicity specific features can enable tailored therapeutics research and improve our understanding of the interplay between age, ethnicity, and metabolism in global populations. OBJECTIVE: A metabolomics approach was adopted to profile serum metabolomes in middle-aged and elderly men of different ethnicities from the Northwest of England, UK. METHODS: Serum samples from 572 men of White European (WE), South Asian (SA), and African-Caribbean (AC) ethnicities, ranging between 40 and 86 years were analysed. A combination of liquid chromatography (LC) and gas chromatography (GC) coupled to high-resolution mass spectrometry (MS) was used to generate the metabolomic profiles. Partial Least Squares Discriminant Analysis (PLS-DA) based classification models were built and validated using resampling via bootstrap analysis and permutation testing. Features were putatively annotated using public Human Metabolome Database (HMDB) and Golm Metabolite Database (GMD). Variable Importance in Projection (VIP) scores were used to determine features of interest, after which pathway enrichment analysis was performed. RESULTS: Using profiles from our analysis we classify subjects by their ethnicity with an average correct classification rate (CCR) of 90.53% (LC-MS data) and 85.58% (GC-MS data). Similar classification by age (< 60 vs. ≥ 60 years) returned CCRs of 90.20% (LC-MS) and 71.13% (GC-MS). VIP scores driven feature selection revealed important compounds from putatively annotated lipids (subclasses including fatty acids and carboxylic acids, glycerophospholipids, steroids), organic acids, amino acid derivatives as key contributors to the classifications. Pathway enrichment analysis using these features revealed statistically significant perturbations in energy metabolism (TCA cycle), N-Glycan and unsaturated fatty acid biosynthesis linked pathways amongst others. CONCLUSION: We report metabolic differences measured in serum that can be attributed to ethnicity and age in healthy population. These results strongly emphasise the need to consider confounding effects of inherent metabolic variations driven by ethnicity of participants in population-based metabolic profiling studies. Interpretation of energy metabolism, N-Glycan and fatty acid biosynthesis should be carefully decoupled from the underlying differences in ethnicity of participants.