Exploratory analysis of one versus two-day intermittent fasting protocols on the gut microbiome and plasma metabolome in adults with overweight/obesity

Abstract

Nutritional interventions are a promising therapeutic option for addressing obesity and cardiometabolic dysfunction. One such option, intermittent fasting (IF), has emerged as a viable alternative to daily caloric restriction and may beneficially modulate body weight regulation and alter the gut microbiome (GM) and plasma metabolome. This secondary analysis of a larger, registered trial (ClinicalTrials.gov ID: NCT04327141) examined the effect of a four-week intervention comparing one vs. two-consecutive days of IF in combination with protein pacing (IF-P; 4-5 meals/day, >30% protein/day) on the GM, the plasma metabolome, and associated clinical outcomes in overweight and obese adults. Participants (n = 20) were randomly assigned to either a diet consisting of one fasting day (total of 36 h) and six low-calorie P days per week (IF1-P, n = 10) or two fasting days (60 h total) and five low-calorie P days per week (IF2-P, n = 10). The fecal microbiome, clinical outcomes, and plasma metabolome were analyzed at baseline (week 0) and after four weeks. There were no significant time or interaction effects for alpha diversity; however, baseline alpha diversity was negatively correlated with percent body fat change after the four-week intervention (p = 0.030). In addition, beta-diversity for both IF groups was altered significantly by time (p = 0.001), with no significant differences between groups. The IF1-P group had a significant increase in abundance of Ruminococcaceae Incertae Sedis and Eubacterium fissicatena group (q ≤ 0.007), while the IF2-P group had a significant increase in abundance of Ruminococcaceae Incertae Sedis and a decrease in Eubacterium ventriosum group (q ≤ 0.005). The plasma metabolite profile of IF2-P participants displayed significant increases in serine, trimethylamine oxide (TMAO), levulinic acid, 3-aminobutyric acid, citrate, isocitrate, and glucuronic acid (q ≤ 0.049) compared to IF1-P. Fecal short-chain fatty acid concentrations did not differ significantly by time or between groups (p ≥ 0.126). Interestingly, gastrointestinal symptoms were significantly reduced for the IF2-P group but not for the IF1-P group. Our results demonstrate that short-term IF modestly influenced the GM community structure and the plasma metabolome, suggesting these protocols could be viable for certain nutritional intervention strategies.

Keywords: caloric restriction; gastrointestinal symptoms; gut microbiome; intermittent fasting; metabolome; obesity; protein pacing; weight loss.

Conflict of interest statement

This study received funding from Isagenix International, LLC and they were not involved in the study design, collection, analysis, interpretation of data, and the writing of this article or the decision to submit it for publication. Authors AM and EG were employed by Isagenix International, LLC and contributed to the study as noted above. AM was a doctoral candidate at Arizona State University, and this work was in partial fulfillment of his dissertation. The study was conducted at Skidmore College and AM was blinded throughout data collection and analyses. Author PA is a member of the scientific advisory board at Isagenix International, LLC and the International Protein Board. Author CW is on the scientific advisory board for the Wheat Foods Council and the Hass Avocado Board’s Avocado Nutrition Science Advisory. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2022 Mohr, Jasbi, Bowes, Dirks, Whisner, Arciero, Poe, Gu, Gumpricht, Sweazea and Arciero.

Figures

FIGURE 1

Overview of study design. After enrollment, a one-week run-in period, and randomization, 20 overweight and obese males and females followed an intermittent fasting, protein pacing-based weight loss diet consisting of one (IF1-P, n = 10) or two (IF2-P, n = 10) fasting days per week for four weeks equated for weekly energy and macronutrient intake. Clinical data and fecal and blood samples were collected at baseline and after week four. “Created with BioRender.com”.

FIGURE 2

Variation in gut microbiome diversity metrics at baseline and week four of IF1-P and IF2-P participants. (A) Non-metric multidimensional scaling (NMDS) ordination of Bray-Curtis dissimilarity matrix. The GM of both IF1-P (n = 10) and IF2-P (n = 10) groups shifted significantly from baseline to week four (R2 = 0.042, p < 0.001), but there was no difference between groups by time (R2 = 0.009, p = 0.823). The same participant is connected by a dotted line, starting at baseline and the arrow pointing to the end of the intervention period. (B) The first distances of the Bray-Curtis dissimilarities between IF1-P and IF2-P were not significant (p = 0.579). Boxes denote the interquartile range (IQR) between the first and third quartiles, and the horizontal line defines the median. (C) First principal coordinate (PCo1) values differed by time (p = 0.013), with no differences between groups over time detected (p = 0.473). (D) Shannon diversity index did not change significantly over time for IF1-P and IF2-P groups (p ≥ 0.341). (E) Faith’s PD diversity index did not change significantly over time for IF1-P and IF2-P groups (p = 0.653). Boxes denote the IQR between the first and third quartiles, and the horizontal line defines the median. A dotted line connects the same participant. (F) Correlation of baseline Shannon Diversity with percent body fat loss from the four-week IF intervention. (G) Correlation of baseline Faith’s phylogenetic diversity (PD) with percent body fat loss from the four-week IF intervention. Both groups were combined for the correlation analyses and are displayed in different colors. The gray cloud around the regression line indicates the 95% confidence interval.

FIGURE 3

(A) Total bacterial number observed in both IF1-P (n = 10) and IF2-P (n = 10) groups at baseline and week four. No significant effects of time or interaction (p ≥ 0.603) were noted. Total bacterial numbers were calculated as average copies of 16S rRNA gene/g wet feces via qPCR. Group means at each time point are displayed as black bars. (B) Average relative abundance of the most prevalent gut microbiome phyla among study participants for IF1-P (n = 10) and IF2-P (n = 10) at baseline and week four by 16S rRNA sequencing. Phyla with a median relative abundance of less than 1% are collapsed into the category “Other”. (C) Average relative abundance of the most prevalent gut microbiome genera among study participants for IF1-P (n = 10) and IF2-P (n = 10) at baseline and week four by 16S rRNA sequencing. Genera with a median relative abundance of less than 1% are collapsed into the category “Other”.

FIGURE 4

(A) Genera were differentially abundant between baseline and week four for IF1-P participants (n = 10). (B) Genera that were differentially abundant between baseline and week four for IF2-P participants (n = 10). Points represent each genera’s log2 fold change (log2FC; effect size). A positive value indicates that a feature increased in abundance at week four (right), and a negative value indicates a decrease in abundance at week four (left). Bars represent 95% confidence intervals derived from the ANCOM-BC model. Note, the genus Incertae Sedis is from the Ruminococcaceae family. Boxplots displaying statistically significant differences in (C) peptidoglycan biosynthesis II and (D) chorismate biosynthesis II from baseline to week four of predicted metabolic pathways for IF1-P participants. Boxplot displaying a statistically significant difference in (E) adenosine nucleotides degradation IV from baseline to week four of the predicted metabolic pathway for IF2-P participants. Pathways are displayed as centered log-ratio (CLR) transformed abundances. Boxes denote the interquartile range (IQR) between the first and third quartiles, and the horizontal line defines the median.

FIGURE 5

(A) GLM adjusted for age, sex, BMI, and time, with FDR-correction, and (B) box plots of significant metabolites as indicated by GLM: serine (q = 0.003), TMAO (q = 0.012), levulinic acid (q = 0.017), 3-aminobutyric acid (q = 0.029), citrate (q = 0.033), isocitrate (q = 0.033), and glucuronic acid (q = 0.049). Red lines in box plots denote optimal cutoff values as calculated by the Youden method, black lines indicate median values, and yellow diamonds show group averages.

FIGURE 6

(A) Scores plot of OPLS-DA model constructed using seven significant metabolites identified by GLM, showing percent variance accounted for by experimental and orthogonal data. (B) OPLS-DA model overview showing predictive and explanatory capacity (R2X = 0.351, R2Y = 0.237, Q2 = 0.202); y-axis represents proportion of total variance. (C) Permutation test with 100 iterations showing model fit distributions (Perm R2Y p < 0.01, Perm Q2 p < 0.01). (D) ROC analysis of OPLS-DA model for assessing IF-P (AUC = 0.83, 95% CI: 0.70-0.94, sensitivity = 0.8, specificity = 0.8). (E) Box plot of OPLS-DA predictive values; the red line in the box plot denotes the optimal cutoff value, while yellow diamonds show group averages, and black lines illustrate group medians.