Scientific Report

\documentclass[12pt,a4paper]{article}

\usepackage[utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage{lmodern}
\usepackage[margin=1in]{geometry}
\usepackage{amsmath}
\usepackage{booktabs}
\usepackage{graphicx}
\usepackage{hyperref}
\usepackage{setspace}
\usepackage{enumitem}
\usepackage{tabularx}
\usepackage{siunitx}

\sisetup{group-separator={,}, group-minimum-digits=4}

\title{\textbf{Effects of Intermittent Fasting on Gut Microbiome Composition and Metabolic Markers in Adults with Prediabetes: A Randomized Controlled Trial}}

\author{
  Rachel M.~Foster$^{1,*}$, David K.~Okonkwo$^{2}$, Yuki Tanaka$^{1}$,\\
  Maria Elena Ruiz$^{3}$, Henrik Johansson$^{4}$, Aisha Patel$^{1}$\\[8pt]
  \small $^{1}$Department of Nutritional Sciences, University of California, Davis, CA, USA\\
  \small $^{2}$Division of Endocrinology, Johns Hopkins University School of Medicine, Baltimore, MD, USA\\
  \small $^{3}$Gut Microbiome Research Center, University of Barcelona, Spain\\
  \small $^{4}$Department of Biostatistics, Karolinska Institute, Stockholm, Sweden\\[6pt]
  \small $^{*}$Corresponding author: [email protected]
}

\date{}

\begin{document}

\maketitle

\begin{abstract}
\noindent\textbf{Background:} Intermittent fasting (IF) has shown promise in improving metabolic health, but its effects on gut microbiome composition and the relationship between microbial changes and metabolic outcomes in prediabetic populations remain poorly understood.

\noindent\textbf{Objective:} To evaluate the effects of two intermittent fasting protocols on gut microbiome diversity, composition, and metabolic markers in adults with prediabetes over 16 weeks.

\noindent\textbf{Methods:} In this three-arm, parallel-group randomized controlled trial, 186 adults with prediabetes (HbA1c 5.7--6.4\%) were randomized to time-restricted eating (TRE; 8-hour eating window), alternate-day fasting (ADF; 500 kcal on fast days), or a control diet (standard dietary guidance). Fecal samples were collected at baseline, week 8, and week 16 for 16S rRNA gene sequencing and shotgun metagenomic analysis. Primary outcomes were changes in gut microbiome alpha diversity (Shannon index) and metabolic markers (HbA1c, fasting glucose, HOMA-IR).

\noindent\textbf{Results:} At 16 weeks, both IF groups showed significant increases in Shannon diversity compared to control (TRE: +0.42 $\pm$ 0.18, $p = 0.003$; ADF: +0.58 $\pm$ 0.21, $p < 0.001$). ADF produced greater improvements in HbA1c ($-0.31\%$ vs.\ $-0.18\%$, $p = 0.024$) and HOMA-IR ($-1.14$ vs.\ $-0.72$, $p = 0.018$) compared to TRE. Mediation analysis revealed that enrichment of \textit{Akkermansia muciniphila} and \textit{Faecalibacterium prausnitzii} mediated 34\% of the effect of ADF on insulin resistance.

\noindent\textbf{Conclusion:} Both intermittent fasting protocols improved gut microbiome diversity and metabolic markers in prediabetic adults, with ADF showing superior effects. Specific microbial taxa partially mediate the metabolic benefits, suggesting the gut microbiome as a mechanistic pathway linking intermittent fasting to improved glucose metabolism.

\noindent\textbf{Keywords:} intermittent fasting, gut microbiome, prediabetes, insulin resistance, time-restricted eating, randomized controlled trial
\end{abstract}

\onehalfspacing

\section{Introduction}

Prediabetes affects approximately 96 million adults in the United States alone, representing a critical intervention window before progression to type 2 diabetes mellitus~\cite{cdc2022}. While lifestyle modifications remain the cornerstone of prediabetes management, conventional dietary interventions achieve modest and often unsustained results, with only 5--10\% of individuals maintaining long-term improvements in glycemic control~\cite{knowler2002}.

Intermittent fasting (IF) has emerged as an alternative dietary strategy with growing evidence of metabolic benefits. Two predominant IF protocols---time-restricted eating (TRE), which limits daily food intake to a defined window (typically 6--10 hours), and alternate-day fasting (ADF), which alternates between normal eating and severe caloric restriction---have demonstrated improvements in body weight, insulin sensitivity, and inflammatory markers in short-term studies~\cite{varady2021}. However, the mechanisms underlying these benefits remain incompletely understood.

The gut microbiome has increasingly been recognized as a key mediator of metabolic health. Alterations in microbial composition and diversity have been associated with insulin resistance, obesity, and type 2 diabetes~\cite{qin2012}. Preclinical studies suggest that fasting-induced changes in the gut microbiome, including increased abundance of beneficial taxa such as \textit{Akkermansia muciniphila}, may contribute to improved metabolic outcomes~\cite{li2017}. However, human data linking IF-induced microbiome changes to metabolic improvements in at-risk populations are scarce.

This randomized controlled trial was designed to address three questions: (1) Do TRE and ADF differentially affect gut microbiome composition and diversity? (2) Do the two IF protocols produce different metabolic outcomes in adults with prediabetes? (3) To what extent do microbiome changes mediate the metabolic effects of IF?

\section{Materials and Methods}

\subsection{Study Design and Participants}

This was a 16-week, three-arm, parallel-group, single-blind (outcome assessors) randomized controlled trial conducted at the UC Davis Clinical Research Center between March 2025 and January 2026. The study was approved by the UC Davis Institutional Review Board (Protocol \#2024-1847) and registered at ClinicalTrials.gov (NCT05XXX789). All participants provided written informed consent.

Eligible participants were adults aged 30--65 years with prediabetes (HbA1c 5.7--6.4\% or fasting plasma glucose 100--125 mg/dL), body mass index (BMI) 25.0--39.9 kg/m$^2$, and stable body weight ($\pm$3 kg) for at least 3 months prior to enrollment. Exclusion criteria included current use of diabetes medications, antibiotics within 3 months, probiotic supplements within 1 month, history of eating disorders, pregnancy or lactation, and significant gastrointestinal, hepatic, or renal disease.

\subsection{Randomization and Interventions}

Participants were randomized 1:1:1 using computer-generated block randomization (block size 6), stratified by sex and BMI category ($<$30 vs.\ $\geq$30 kg/m$^2$), to one of three groups:

\begin{enumerate}[leftmargin=*]
  \item \textbf{Time-Restricted Eating (TRE):} Participants consumed all daily calories within an 8-hour window (10:00 AM--6:00 PM), with only water, black coffee, or unsweetened tea permitted outside this window. No specific caloric or macronutrient targets were prescribed.
  \item \textbf{Alternate-Day Fasting (ADF):} Participants alternated between ad libitum eating days and fast days (500 kcal for women, 600 kcal for men, consumed as a single meal between 12:00--2:00 PM). Fast-day meals were provided by the study kitchen.
  \item \textbf{Control:} Participants received standard dietary counseling based on the Dietary Guidelines for Americans with no specific meal timing instructions.
\end{enumerate}

All participants attended biweekly check-ins with a registered dietitian and completed daily dietary logs via a mobile application.

\subsection{Outcome Measures}

The co-primary outcomes were change in gut microbiome alpha diversity (Shannon index) and change in HbA1c from baseline to week 16. Secondary outcomes included fasting plasma glucose, fasting insulin, HOMA-IR, body weight, waist circumference, lipid panel, and inflammatory markers (hs-CRP, IL-6, TNF-$\alpha$).

\subsection{Microbiome Analysis}

Fecal samples were collected at baseline, week 8, and week 16 using Zymo DNA/RNA Shield Collection Kits. Genomic DNA was extracted using the MoBio PowerSoil kit. 16S rRNA gene sequencing (V3--V4 region) was performed on the Illumina MiSeq platform, and a subset of 90 samples (30 per group at baseline and week 16) underwent shotgun metagenomic sequencing on the Illumina NovaSeq 6000.

Amplicon sequence variants (ASVs) were generated using DADA2. Taxonomic classification used the SILVA v138 reference database. Alpha diversity was calculated using the Shannon and Chao1 indices. Beta diversity was assessed using weighted UniFrac distances and visualized with principal coordinates analysis (PCoA). Differential abundance analysis was performed using DESeq2 with Benjamini--Hochberg correction for multiple comparisons.

\subsection{Statistical Analysis}

Sample size was calculated to detect a between-group difference of 0.35 in Shannon index change (SD = 0.50, $\alpha$ = 0.05, power = 0.80), requiring 54 participants per group. Accounting for 15\% dropout, we targeted 63 per group.

Primary analyses used linear mixed-effects models with group, time, and group$\times$time interaction as fixed effects and participant as a random effect (intention-to-treat population). Mediation analysis was conducted using the Baron and Kenny framework with bootstrapped confidence intervals (5,000 iterations). All analyses were performed in R version 4.3.2. Two-sided $p$-values $< 0.05$ were considered statistically significant.

\section{Results}

\subsection{Participant Flow and Baseline Characteristics}

Of 312 individuals screened, 186 were randomized (TRE: $n = 62$; ADF: $n = 62$; Control: $n = 62$). Twenty-two participants (11.8\%) withdrew before week 16 (TRE: 6; ADF: 9; Control: 7), with no significant difference in dropout rates between groups ($p = 0.61$). The intention-to-treat analysis included all 186 randomized participants.

\begin{table}[h]
\centering
\caption{Baseline characteristics of study participants.}
\label{tab:baseline}
\begin{tabular}{lccc}
\toprule
\textbf{Characteristic} & \textbf{TRE ($n=62$)} & \textbf{ADF ($n=62$)} & \textbf{Control ($n=62$)} \\
\midrule
Age, years (mean $\pm$ SD) & 48.3 $\pm$ 9.1 & 47.8 $\pm$ 8.7 & 49.1 $\pm$ 9.4 \\
Female sex, $n$ (\%) & 36 (58.1) & 34 (54.8) & 37 (59.7) \\
BMI, kg/m$^2$ & 31.4 $\pm$ 3.8 & 31.8 $\pm$ 4.1 & 31.1 $\pm$ 3.6 \\
HbA1c, \% & 5.92 $\pm$ 0.21 & 5.95 $\pm$ 0.19 & 5.90 $\pm$ 0.22 \\
Fasting glucose, mg/dL & 108.4 $\pm$ 7.2 & 109.1 $\pm$ 6.8 & 107.8 $\pm$ 7.5 \\
HOMA-IR & 3.82 $\pm$ 1.14 & 3.91 $\pm$ 1.22 & 3.77 $\pm$ 1.08 \\
Shannon diversity index & 3.24 $\pm$ 0.48 & 3.18 $\pm$ 0.52 & 3.21 $\pm$ 0.46 \\
\bottomrule
\end{tabular}
\end{table}

Baseline characteristics were well balanced across groups (Table~\ref{tab:baseline}).

\subsection{Gut Microbiome Changes}

Both IF groups showed significant increases in alpha diversity at week 16 (Table~\ref{tab:results}). The ADF group demonstrated the largest increase in Shannon index (+0.58 $\pm$ 0.21, $p < 0.001$ vs.\ control), followed by TRE (+0.42 $\pm$ 0.18, $p = 0.003$). The difference between ADF and TRE was statistically significant ($p = 0.031$).

Beta diversity analysis revealed significant compositional shifts in both IF groups compared to control (PERMANOVA: TRE, $R^2 = 0.034$, $p = 0.008$; ADF, $R^2 = 0.051$, $p < 0.001$).

At the genus level, the most prominent changes in the ADF group included increased relative abundance of \textit{Akkermansia} (2.1\% to 5.8\%, $q < 0.001$), \textit{Faecalibacterium} (8.4\% to 12.7\%, $q = 0.003$), and \textit{Bifidobacterium} (3.2\% to 5.1\%, $q = 0.012$), alongside decreased \textit{Bacteroides} (24.8\% to 19.3\%, $q = 0.008$). TRE showed similar but attenuated changes.

\subsection{Metabolic Outcomes}

\begin{table}[h]
\centering
\caption{Changes in primary and secondary outcomes at week 16 (mean $\pm$ SD).}
\label{tab:results}
\small
\begin{tabular}{lcccc}
\toprule
\textbf{Outcome} & \textbf{TRE} & \textbf{ADF} & \textbf{Control} & \textbf{$p$ (ANOVA)} \\
\midrule
$\Delta$ Shannon index & +0.42 $\pm$ 0.18 & +0.58 $\pm$ 0.21 & +0.08 $\pm$ 0.15 & $<$0.001 \\
$\Delta$ HbA1c (\%) & $-$0.18 $\pm$ 0.12 & $-$0.31 $\pm$ 0.15 & $-$0.04 $\pm$ 0.10 & $<$0.001 \\
$\Delta$ Fasting glucose (mg/dL) & $-$6.2 $\pm$ 4.8 & $-$10.4 $\pm$ 5.3 & $-$1.8 $\pm$ 3.9 & $<$0.001 \\
$\Delta$ HOMA-IR & $-$0.72 $\pm$ 0.48 & $-$1.14 $\pm$ 0.56 & $-$0.18 $\pm$ 0.34 & $<$0.001 \\
$\Delta$ Body weight (kg) & $-$3.1 $\pm$ 2.4 & $-$4.8 $\pm$ 2.7 & $-$0.9 $\pm$ 1.8 & $<$0.001 \\
$\Delta$ Waist circumference (cm) & $-$2.8 $\pm$ 1.9 & $-$4.2 $\pm$ 2.3 & $-$0.6 $\pm$ 1.4 & $<$0.001 \\
$\Delta$ hs-CRP (mg/L) & $-$0.84 $\pm$ 0.62 & $-$1.21 $\pm$ 0.74 & $-$0.12 $\pm$ 0.48 & $<$0.001 \\
$\Delta$ Triglycerides (mg/dL) & $-$18.4 $\pm$ 14.2 & $-$26.7 $\pm$ 16.8 & $-$4.2 $\pm$ 11.3 & $<$0.001 \\
\bottomrule
\end{tabular}
\end{table}

ADF produced significantly greater reductions in HbA1c compared to TRE ($p = 0.024$) and control ($p < 0.001$). Similar patterns were observed for HOMA-IR, fasting glucose, and inflammatory markers (Table~\ref{tab:results}).

\subsection{Mediation Analysis}

Mediation analysis identified changes in \textit{Akkermansia muciniphila} abundance as a significant mediator of the ADF effect on HOMA-IR (indirect effect: $-0.39$, 95\% CI: $-0.58$ to $-0.18$), accounting for 34.2\% of the total effect. \textit{Faecalibacterium prausnitzii} mediated an additional 12.8\% (indirect effect: $-0.15$, 95\% CI: $-0.28$ to $-0.04$). The combined microbial mediation was 47.0\% (95\% CI: 31.4\% to 62.8\%).

\section{Discussion}

This randomized controlled trial provides evidence that intermittent fasting, particularly alternate-day fasting, produces meaningful improvements in both gut microbiome composition and metabolic markers in adults with prediabetes over 16 weeks. Notably, our mediation analysis suggests that microbiome changes are not merely epiphenomenal but may mechanistically contribute to the metabolic benefits of IF.

The enrichment of \textit{A.\ muciniphila} observed in both IF groups is consistent with preclinical data showing that caloric restriction promotes this mucin-degrading bacterium~\cite{li2017}. \textit{A.\ muciniphila} has been associated with improved intestinal barrier function, reduced metabolic endotoxemia, and enhanced insulin sensitivity in both animal models and human studies. The finding that this taxon mediates over one-third of the ADF effect on insulin resistance provides the strongest human evidence to date for a causal pathway linking fasting, microbiome changes, and metabolic improvement.

Our study has several limitations. First, the 16-week duration may be insufficient to assess long-term sustainability and clinical endpoints. Second, despite providing fast-day meals to the ADF group, dietary compliance relied partly on self-report. Third, our predominantly urban, university-affiliated sample may limit generalizability. Fourth, while mediation analysis suggests mechanistic pathways, causal inference from observational within-trial analyses requires confirmation in dedicated mechanistic studies.

\section{Conclusion}

Both time-restricted eating and alternate-day fasting improve gut microbiome diversity and metabolic markers in adults with prediabetes, with ADF demonstrating superior effects. Enrichment of specific beneficial taxa, particularly \textit{Akkermansia muciniphila}, partially mediates the metabolic benefits. These findings support the gut microbiome as a therapeutic target in prediabetes management and suggest that IF protocols may be optimized by considering their differential effects on microbial ecology.

\section*{Acknowledgements}

The authors thank the study participants for their commitment, the UC Davis Clinical Research Center staff, and the UC Davis Genome Center for sequencing support. We gratefully acknowledge statistical consultation from Dr.\ Lisa Wong.

\section*{Funding}

This work was supported by NIH/NIDDK grant R01 DK123456, the UC Davis Clinical and Translational Science Center (UL1 TR001860), and the Swedish Research Council grant 2023-04521.

\section*{Conflict of Interest}

The authors declare no competing interests.

\bibliographystyle{unsrt}
\begin{thebibliography}{9}
\bibitem{cdc2022} Centers for Disease Control and Prevention, ``National Diabetes Statistics Report 2022,'' Atlanta, GA: U.S.\ Dept.\ of Health and Human Services, 2022.
\bibitem{knowler2002} W.~C.~Knowler \textit{et al.}, ``Reduction in the incidence of type 2 diabetes with lifestyle intervention or metformin,'' \textit{N Engl J Med}, vol.~346, pp.~393--403, 2002.
\bibitem{varady2021} K.~A.~Varady \textit{et al.}, ``Clinical application of intermittent fasting for weight loss: progress and future directions,'' \textit{Nat Rev Endocrinol}, vol.~18, pp.~309--321, 2022.
\bibitem{qin2012} J.~Qin \textit{et al.}, ``A metagenome-wide association study of gut microbiota in type 2 diabetes,'' \textit{Nature}, vol.~490, pp.~55--60, 2012.
\bibitem{li2017} G.~Li \textit{et al.}, ``Intermittent fasting promotes white adipose browning and decreases obesity by shaping the gut microbiota,'' \textit{Cell Metab}, vol.~26, pp.~672--685, 2017.
\end{thebibliography}

\end{document}