Fig. 1: Descriptive analysis of participant and metabolite data.

a Distribution of follow-up times to diagnosis. b Relationship between reported alcohol intake and plasma ethanol concentrations. c Plasma metabolites identified and their correlations. No significant association was found between quartile of alcohol intake and plasma ethanol concentration (Kruskal–Wallis P > 0.05 for pre- and postmenopausal groups). Correlations were calculated from fasted participants only and metabolites are ordered by hierarchical cluster as determined by Ward’s method.