Gut microbiome in healthy adults during prolonged fasting

This study examines the metabolic and microbiome effects of prolonged fasting in healthy adults, employing a randomized, waitlist-controlled design. Thirty-eight participants completed a 5-day fasting intervention, followed by a 12-week follow-up. Clinical, microbiome, and metabolomic profiling were performed at multiple time points to assess physiological and microbial responses. Fasting significantly reduced resting energy expenditure (REE), body mass index (BMI), and fat mass. Although many parameters returned to baseline after 12 weeks, sustained BMI reduction was observed in a subset of participants.</p><p>Shotgun metagenomic sequencing of fecal samples revealed pronounced but reversible shifts in gut microbiota composition and function. Microbiome alpha diversity changes were associated with individual baseline diversity. Nuclear magnetic resonance (NMR) metabolomics of plasma and feces revealed short-term alterations in energy and lipid metabolism, including elevated ketone body levels and reduced glucose levels, which largely returned to baseline during follow-up.</p><p>To explore individual variability in fasting response, a machine learning model was developed to predict long-term BMI reduction based on baseline microbiome and clinical data. The final model identified Faecalibacterium sp., Oscillibacter sp. 57_20, LDL cholesterol, and systolic blood pressure as key predictors of weight loss response. This model was validated in independent cohorts of patients with metabolic syndrome and multiple sclerosis, demonstrating accurate cross-population performance.</p><p>The study provides mechanistic insight into the interplay between gut microbiota and host metabolic adaptation to fasting. It highlights that predictive models leveraging microbiome and clinical parameters may identify individuals most likely to benefit from dietary interventions. Fasting-induced microbiome and metabolome shifts were largely reversible, but selected microbial taxa and metabolites, including uridine, butyrate, and tryptophan-related compounds, were associated with sustained weight loss. These findings suggest a role for gut-brain-metabolism interactions in mediating fasting outcomes.</p><p>Fecal samples were collected at multiple study visits and processed for both metagenomic and metabolomic analyses. DNA was extracted using the ZymoBIOMICS kit and sequenced via Illumina NovaSeq 6000 (150 bp paired-end). Sequencing reads were mapped to the mOTUv2 database, and data were rarefied to standardize read depth. Plasma and fecal metabolites were analyzed by high-resolution NMR spectroscopy. Statistical analysis included mixed models, PERMANOVA, and permutation-based correlation testing. Machine learning modeling was applied using random forest classification with leave-one-out cross-validation and feature selection via Boruta and recursive elimination.</p><p>This study was conducted at the Experimental and Clinical Research Center, Berlin, Germany, with all participants providing informed consent. The trial is registered at ClinicalTrials.gov (NCT04452916). The dataset enables exploration of microbiome-mediated metabolic adaptation and supports the development of precision nutrition approaches informed by baseline microbiome composition.

Classified as human gut / stool shotgun DNA metagenome of adult by sonnet_x2 with confidence 0.85. Assay shotgun_dna.

• host study.description · “prolonged fasting in healthy adults”
• body site study.description · “Shotgun metagenomic sequencing of fecal samples”
• life stage study.description · “prolonged fasting in healthy adults”

98 runs over 98 samples · mean 7.82 Gbp per run (median 7.64 Gbp, total 766.35 Gbp; 98 runs with a base count) · layout PAIRED 98 · platform ILLUMINA 98 · models Illumina NovaSeq 6000 (98) · 0% of runs with a Sandpiper profile · first public 2026-09-01

Kaufhold G first · Bartolomaeus TUP · Schütte K · Schütte T · Kamboj S · Löber U · Rahn G · McParland V · Braun L · Markó L · Mammadli M · Krannich A · Bahr LS · Gutmann F · Paul F · Ducarmon QR · Zeller G · Mesnage R · Wilck N · Zernecke A · Oefner PJ · Gronwald W · Müller DN · Forslund-Startceva SK · Bähring S · Bartolomaeus H · Siebert N last

PRJNA1293883.csv.gz (3.9 KB) · parquet (76.8 KB) — all 98 samples, every field with route and confidence · PRJNA1293883_determinations.csv.gz (10.5 KB) — 1,274 evidence rows