Skip to main content
. 2022 Dec 20;117(2):286–297. doi: 10.1016/j.ajcnut.2022.10.015

Figure 3.

Figure 3

Prolonged fasting dramatically alters human metabolome and upregulates bioactive metabolites. Volcano plot analysis of the fasted metabolite levels compared with the baseline metabolite levels showing over 375 significantly different metabolites between the 2 states (A). Principal component analysis of complete metabolite datasets between the baseline, fed, fasted, and re-fed states (B). Circulating levels of the ketone body beta-hydroxybutyrate (BHB) in participant plasma (C). Significance between timepoints for all metabolite levels (C–E, G–J) is denoted by their assigned letter (a–c) where groups that share an assigned letter in common were found to have no significant differences (P > 0.05) and groups that do not share an assigned letter in common were found to be significantly different (P < 0.05) from each other. The significance levels are provided for the baseline versus fasted states and fed versus re-fed states where significance (P < 0.05) is achieved. Baseline versus fasted (1.16 ± 1.36 vs. 19.80 ± 10.80 RA, P < 0.0001). Circulating levels of (R)-3-hydroxybutyrylcarnitine in participant plasma (D). Baseline versus fasted states (0.57 ± 0.45 vs. 7.98 ± 5.10 RA, P < 0.0001) and fed versus re-fed states (0.60 ± 0.60 vs. 1.59 ± 1.34, P < 0.0001). Circulating levels of the ketone body acetoacetate in participant plasma (E). Baseline versus fasted states (0.75 ± 0.86 vs. 9.89 ± 5.37, P < 0.0001) and fed versus re-fed states (0.78 ± 0.63 vs. 1.52 ± 1.32, P = 0.025). Pathway analysis of metabolic datasets showing significantly differentially regulated pathways between the baseline and fasted states determined using Fisher’s exact test (F). Circulating levels of spermidine in participant plasma (G). Baseline versus fasted states (1.03 ± 0.37 vs. 1.51 ± 0.35, P = 0.0001) and fed versus re-fed states (0.74 ± 0.37 vs. 0.87 ± 0.27, P = 0.025). Circulating levels of 1-methylnicotinamide (1-MNA) in participant plasma (H). Baseline versus fasted states (1.52 ± 0.84 vs. 2.28 ± 1.19, P = 0.0038). Circulating levels of palmitoylethanolamide (PEA) in participant plasma (I). Baseline versus fasted states (0.87 ± 0.30 vs. 2.06 ± 0.51, P < 0.0001). Circulating levels of oleoylethanolamide (OEA) in participant plasma (J). Baseline versus fasted states (0.93 ± 0.46 vs. 1.83 ± 0.61, P < 0.0001). The n of each analysis described earlier was 20. Significance for Figure A–E and G–J was determined using a linear fit model. Moderated t-statistics were computed by empirical Bayes moderation, and P values were adjusted using the Benjamini–Hochberg method. Significance was determined using Fisher’s exact test for Figure F. Boxplots displayed on graphs show the median, first quartile, third quartile, minimum, and maximal values of each timepoint.