Metabolomics of Coffee Consumption

Coffee is among the most widely consumed beverages in the world and has received considerable attention regarding health risks and benefits. Coffee is a major source of caffeine, but it also contains many other bioactive compounds, any of which may impact pathways related to disease development. Population‐based metabolomic studies observe correlations between habitual coffee consumption and plasma or urine levels of known coffee derivatives but also other novel metabolites. Clinical metabolomic applications have targeted only coffee‐derived metabolites and following very recent coffee intake. The objective of this study was to conduct a comprehensive metabolomic study of controlled coffee intake. We examined plasma samples collected from a previously conducted 3‐stage coffee trial (Kempf et al, AJCN, 2010). During the first month, participants (N=47) refrained from drinking coffee, whereas in the second month they were instructed to consume 4 cups of filtered coffee/d and in the third month 8 cups/d. Fasting plasma samples collected after each coffee stage were subject to nontargeted metabolomic profiling using UPLC‐ESI‐MS/MS. After quality control, 732 metabolites were included for univariate and multivariate analysis which included repeated measures ANOVA (RMA) and multilevel sparse partial least squares discriminant analysis (msPLSDA), respectively. RMA revealed 115 metabolites significantly associated with coffee intake (P<0.05 and Q<0.05). msPLSDA supported clear distinctions between all coffee periods. Upon evaluation of the performance of the fitted msPLSDA model, consisting of 50 discriminative metabolites in total from the first two principal components, all samples following the non‐coffee period were correctly classified as such (0% error rate). Classification error rates for samples following the 4 cup and 8 cup periods ranged from 9 to 11%. All but 4 metabolites selected as discriminative by msPLSDA were among those identified in RMA. Of the total 119 significant metabolite markers of coffee intake, 82 are of known identity and map to one of 33 pre‐defined biological pathways (Table). Based on RMA results, we observed a significant enrichment of metabolite members of 5 pathways (P<0.05, v): (1) xanthine metabolism‐includes caffeine metabolites, (2) benzoate metabolism‐reflects polyphenol metabolite products of gut‐microbiota metabolism,(3) steroid‐novel but may reflect sterol content of coffee, (4) fatty acid metabolism (acyl choline)‐novel link to coffee and (5) endocannabinoid‐novel link to coffee. Strong correlations of coffee‐induced changes in plasma levels of metabolites within and between these pathways were observed (Fig 1). Another 37 metabolites are of unknown identity but correlations with known metabolites observed for a subset of these may provide insight to their identity. Unsupervised hierarchical clustering analysis of all 119 metabolite markers of coffee intake demonstrates good classification (Fig 2). Excluding the full set of xanthines did not markedly alter classification performance. In conclusion, our study provides the first comprehensive analysis of the metabolomic changes in response to controlled coffee consumption. The novel metabolites and candidate pathways we have identified may provide new insight to the mechanisms by which coffee may be exerting its health effects.Support or Funding InformationThis work was funded by the American Diabetes Association (7–13‐JF‐15 to MCC)