Effect of gut microbiome modulation on muscle function and cognition: the PROMOTE randomised controlled trial

The study addresses whether modulating the gut microbiome with a prebiotic can enhance muscle function and cognition in older adults, beyond the benefits of protein (BCAA) supplementation and resistance exercise. With global population ageing, muscle loss (sarcopenia) and cognitive decline are increasing. Older adults often have lower protein intake and exhibit anabolic resistance of skeletal muscle, attenuating muscle protein synthesis responses to nutrition and exercise. Emerging evidence implicates the gut microbiome in muscle metabolism (gut–muscle axis) and cognitive function (gut–brain axis), with age-related microbiome changes in diversity and resilience. Prebiotics selectively stimulate beneficial gut microbes and have shown benefits for frailty components and cognition in prior studies. No previous trials tested a combined protein plus prebiotic intervention on both physical and cognitive function in older adults, nor specifically leveraged a twin design to control for shared genetic/environmental factors. The primary hypothesis was that adding a prebiotic to BCAA plus resistance exercise improves muscle strength (5× chair rise time) compared with placebo; secondary hypotheses included improvements in cognition, grip strength, physical performance, appetite, and gut microbiome measures.

Prior work shows protein and resistance exercise synergistically improve muscle function in ageing, yet anabolic resistance limits responsiveness. The gut microbiome influences host metabolism and may contribute to anabolic resistance via effects on protein digestion/absorption, gut barrier function, inflammation, and catabolic pathways. Prebiotic interventions in older adults have improved frailty measures (e.g., grip strength, exhaustion) and overall frailty index in some studies; a growing literature supports gut–brain interactions, with preliminary evidence that prebiotics can benefit aspects of cognition and affect. Microbiome composition is shaped by genetics and early environment; twin studies show heritable taxa and increased microbiota similarity within twin pairs, supporting within-pair randomisation to control confounding. No existing studies combined prebiotics with protein supplementation and resistance exercise to assess both muscle and cognitive outcomes in older people or used a twin RCT design.

Design: PROMOTe was a double-blind, placebo-controlled, randomised controlled trial in older same-sex twin pairs (monozygotic or dizygotic), randomised within twin pairs (one to prebiotic, one to placebo) for 12 weeks. Both arms received daily BCAA supplementation and were advised to undertake simple resistance exercises at least twice weekly. The trial was delivered fully remotely using video visits, online questionnaires and cognitive testing, mailed equipment, and postal biological sample collection. Randomisation and blinding: Allocation conducted by King’s Clinical Trials Unit using fixed blocks of two (within twin pairs). Supplements were provided in indistinguishable sachets, labelled only by randomisation code. Participants and researchers remained blinded until analyses were complete. Interventions: All sachets contained 3.32 g BCAA powder (L-leucine 1660 mg, L-isoleucine 830 mg, L-valine 830 mg) daily. Intervention arm sachets additionally contained 7.5 g prebiotic (Darmocare Pre; inulin and fructo-oligosaccharides). Placebo sachets contained 7.5 g maltodextrin. Participants mixed one sachet per day with water or other drink. All participants received written guidance on home-based strength exercises (e.g., squats, sit-to-stand, leg raises, wall press-ups, biceps curls) with weekly text reminders. Participants: Recruited from TwinsUK registry. Eligibility: age ≥60 years; previously reported protein intake <1 g/kg/day; access to a computer/tablet for remote visits. Key exclusions: severe food allergy; antibiotic, protein supplement, prebiotic or probiotic use within prior 3 months; CKD stage ≥3; ≥5% weight loss over 12 months; significant injury/surgery affecting function; concurrent interventional study participation. Outcomes and measurements: Primary outcome was change in 5× chair rise time from baseline to 12 weeks. Secondary outcomes: cognitive factor score (from CANTAB battery: One Touch Stockings of Cambridge, spatial working memory, spatial span, pattern recognition memory, paired associates learning), grip strength, short physical performance battery (SPPB), gait speed, physical activity (IPAQ MET-min/week), appetite (SNAQ), and gut microbiome composition/function. Remote assessments used mailed equipment (dynamometer, measuring ribbon/tape) with real-time video guidance. Dietary intake was recorded using 3-day online food diaries (myfood24) at baseline and study end. Gut microbiome methods: Participants provided stool samples at baseline and study end in DNA/RNA Shield tubes, mailed to the lab. DNA extraction used a customised MagMax Core protocol; shotgun metagenomic sequencing (Illumina 2×150) was performed by Clinical Microbiomics. High-quality non-host reads were mapped to a human gut gene catalogue; metagenomic species (MGS) were annotated against RefSeq and nt databases. Diversity metrics (richness, Shannon, Faith’s), Bray–Curtis and weighted UniFrac distances were computed. Functional profiling used EggNOG-Mapper and KEGG modules. Analyses used relative abundances; rarefaction applied for diversity estimates. Sample size: Based on TwinsUK data, log10 chair rise time SD = 0.126; a 20% relative reduction deemed clinically important. Power calculations indicated 28 per group (56 total) for 80% power; allowing 20% dropouts targeted n=70 (achieved 72). Twin design expected to increase power via close baseline matching. Statistical analysis: Primary and secondary outcomes analyzed with linear mixed-effects models comparing study-end values adjusted for baseline, with twin clustering (family ID and zygosity) as random effects, treatment as fixed effect, and baseline SNAQ included as covariate. Intention-to-treat and per-protocol analyses were performed. Two-sided p<0.05 considered significant. Missing baseline covariates used mean or recent TwinsUK values (pre-specified); outcome missingness assumed missing at random with no imbalance by arm. Dietary intake assessed with paired t-tests and Goldberg cut-offs (adjusted for activity). Microbiome: group comparisons via Wilcoxon signed-rank, PERMANOVA for beta diversity; multiple testing controlled by Benjamini–Hochberg FDR (10%). Associations between microbiota features and changes in chair rise time or cognition used compositional bias-corrected linear models (centred log-ratio transformed deltas), adjusting for arm and twin pair. Twin microbiota similarity tested by Mann–Whitney U and permutation; heritability estimated via ACE model. Trial conduct: Recruitment May–Dec 2021; first visit 17/05/2021; last final visit 20/12/2021. CONSORT flow: 626 assessed; 72 randomised (36 pairs). Prebiotic: 36 allocated, 32 completed; Placebo: 36 allocated, 34 completed. Compliance assessed via self-report and sachet counts; adverse events and medication changes recorded. Trial registered: NCT04309292.

Participants: 72 older adults (36 twin pairs), mean age 73 years (range 63–83), 78% female. Arms were closely matched at baseline. Primary outcome: No significant effect of prebiotic on chair rise time compared with placebo. ITT mixed model: β = 0.579; 95% CI -1.080 to 2.239; p = 0.494. Mean (SD) change: prebiotic 0.88 s (2.16) vs placebo 1.12 s (1.52); p = 0.631. Secondary physical outcomes: No significant differences for grip strength, SPPB, IPAQ MET-minutes, or SNAQ appetite score between arms (all p>0.05). Cognition: Significant improvement in cognitive factor score with prebiotic vs placebo (β = -0.482; 95% CI -0.813 to -0.141; p = 0.014). Paired Associates Learning (PAL) total errors significantly lower in prebiotic vs placebo (β = 7.55; 95% CI 4.65 to 10.46; p = 0.001), indicating better performance. Other individual CANTAB tests not significantly different. Missingness was higher for CANTAB measures without demographic bias in completers. Adverse events and compliance: More mild adverse events in prebiotic (8/36; 22%) vs placebo (2/36; 6%); p=0.041; no serious adverse events. Compliance high and similar between arms (sachet adherence ~79.7% prebiotic vs 77.5% placebo; p=0.37). Dropouts: 2 per arm (6% each) for burden; lost to follow-up: 2 (prebiotic) vs 0 (placebo). Diet: No between-arm differences in fibre intake at baseline or end. Small overall energy intake reduction driven by prebiotic group (-132.4 kcal/day; between-arm difference ~75.6 kcal/day), within measurement error; no significant differences in weight or BMI. Gut microbiome: Robust sequencing (mean 21.6M read pairs/sample; ~82.8% mapping). Twin pairs’ microbiota significantly more similar than unrelated individuals at baseline and end (permutation-validated). Prebiotic vs placebo at study end: 11 significant taxonomic differences, notably increased Actinobacteria/Bifidobacterium and decreased Phocea massiliensis in prebiotic; also lower Anaeromassilibacillus and higher orders Deltaproteobacteria, Lactobacillales, Eubacteriales in prebiotic. No significant alpha or beta diversity differences between arms cross-sectionally, though within prebiotic arm 40 features changed from baseline to end vs 1 in placebo. Alpha diversity declined in prebiotic arm, consistent with selective taxon increases. Microbiota–phenotype associations: Across all participants, improvements in chair rise time correlated with increases in richness (r=-0.347; Padj=0.0159), Shannon (r=-0.250; Padj=0.0486), and Faith’s phylogenetic diversity (r=-0.297; Padj=0.0275). Eight microbiota features correlated with cognition factor score; Actinobacteria at study end associated with better cognition (r=-0.323; Padj=0.0447). Increases in Veillonellaceae/order Veillonellales from baseline to end were positively associated with cognition improvement (r=0.585, Padj≈0.0555; r=0.543, Padj≈0.0772). Study feasibility: Remote conduct was successful with good adherence, demonstrating feasibility of remote trials in older adults.

The trial tested whether adding a prebiotic to standard anabolic stimuli (BCAA supplementation and simple resistance exercises) enhances muscle strength and cognition in older adults while leveraging a twin design to reduce confounding. The prebiotic significantly altered gut microbiota composition, particularly increasing Bifidobacterium, and improved a composite cognitive factor score and PAL test performance compared with placebo. However, it did not improve the primary muscle strength outcome (chair rise time) or other physical function measures over 12 weeks. These findings suggest that prebiotic-mediated modulation of the gut microbiome can beneficially influence cognitive function in ageing, supporting the gut–brain axis hypothesis. The lack of detectable muscle strength benefit may reflect the short intervention duration for muscle remodeling, the choice of chair rise time as a composite proxy rather than a direct isometric strength measure, potential underpowering for a microbiome-targeted effect size, and/or suboptimal prebiotic type or dose for muscle outcomes in a relatively healthy cohort. Microbiome analyses revealed broad taxonomic shifts in the prebiotic group and correlations between improvements in physical performance and increased microbial diversity, as well as associations between cognition and Actinobacteria abundance, consistent with prior links between Bifidobacterium/Actinobacteria and neurocognitive outcomes. The twin design validated significant within-pair microbial similarity and heritability of some responsive taxa (e.g., Bifidobacterium), highlighting genetic/early environmental influences and potentially explaining limited divergence in beta diversity between co-twins over 12 weeks. Operationally, the study shows that remote trial delivery is feasible in older populations with high adherence and minimal serious adverse events, offering a scalable approach to reduce barriers to participation and costs. Overall, the results advance evidence that gut microbiome interventions may ameliorate aspects of cognitive frailty, while indicating that larger, longer, and potentially differently targeted interventions may be needed to impact muscle strength endpoints.

Prebiotic supplementation alongside BCAA and resistance exercise in older twins did not improve chair rise time or other muscle function measures over 12 weeks but did improve cognition relative to placebo and elicited clear, selective gut microbiome shifts (notably increased Bifidobacterium). The trial demonstrates the feasibility and tolerability of a low-cost, remotely delivered microbiome-targeted intervention in older adults. Future research should employ larger sample sizes, longer durations, direct strength assessments (e.g., isometric measures), inclusion of muscle mass measures, exploration of alternative prebiotics/probiotics or dosing strategies, energy-matched placebos, and integrative functional omics (metatranscriptomics/metabolomics) to elucidate mechanisms underpinning gut–muscle and gut–brain axes and to test efficacy on clinically meaningful physical outcomes.

• Duration and power: The 12-week timeframe may be insufficient for muscle remodeling; no prior microbiome-focused trials to inform effect size likely led to underpowering for the muscle primary endpoint.
• Outcome sensitivity: Chair rise time is functionally relevant but a composite proxy; lack of direct isometric strength measures may have limited sensitivity to subtle changes.
• Remote design constraints: Inability to measure muscle mass; potentially less precise remote assessments; higher missingness in CANTAB cognitive tests due to software/device issues.
• Microbiome sampling/analysis: Only two timepoints (baseline/end) without mid-study sampling; functional inference from metagenomics rather than direct metabolomics/transcriptomics.
• Placebo: Maltodextrin not energy-matched to prebiotic; small, non-significant energy intake differences could confound over time.
• Generalisability: Cohort mostly female, healthy volunteers from TwinsUK; results may not generalise to men, younger or more frail populations.
• Digital access: Remote delivery may exclude those with lower digital literacy or access, potentially biasing participation.