Fasting alters the gut microbiome reducing blood pressure and body weight in metabolic syndrome patients

The study addresses how periodic fasting affects the gut microbiome, immune system (immunome), and cardiometabolic outcomes in hypertensive patients with metabolic syndrome (MetS). While Western-style diets promote cardiometabolic disease, dietary interventions such as the DASH diet can reduce risk but often face compliance challenges. The gut microbiota plays a key role in health and disease, with dysbiosis linked to obesity, hypertension, and chronic inflammation. Profound dietary changes, including fasting, can reshape the microbiome and potentially modulate host physiology. This randomized study tests whether adding a 5-day fasting period before a modified DASH diet produces distinct microbiome and immunome changes and improves blood pressure and weight compared to DASH alone, using a multi-omics approach to elucidate mechanisms and predictors of response.

Prior work has shown that Western diets elevate cardiometabolic disease risk and associate with inflammatory microbiome features, whereas Mediterranean-like diets can improve outcomes. Dysbiosis is characteristic of obesity and hypertension, with shifts in Firmicutes/Bacteroidetes and short-chain fatty acid (SCFA) producers implicated. Fasting and dramatic caloric restriction have been reported to impact both host physiology and the microbiome in rodents and small human studies (e.g., Ramadan fasting). Metformin usage is known to produce characteristic functional microbiome signatures distinct from MetS-associated signatures. Mouse studies suggest intermittent fasting can alleviate metabolic impairments, possibly via microbiota changes. However, comprehensive human multi-omics data in MetS examining periodic fasting’s effects on microbiome, immune system, and cardiovascular endpoints have been lacking.

Design: Substudy of a randomized-controlled bi-centric trial (NCT02099968), with blinded outcome assessment. Participants had MetS per NCEP ATP III and systolic hypertension. Randomization used block-randomization stratified by study center and antihypertensive medication intake. Arms and timeline: Two arms—(1) fasting + modified DASH, and (2) modified DASH alone. Visits at baseline (V1), 1 week post-randomization (V2), and 12 weeks (~3 months; V3). Fasting arm underwent two vegan low-calorie days (≤1200 kcal/day) followed by 5 days of 300–350 kcal/day (vegetable juices and broth), then refeeding with a modified, plant-forward DASH diet emphasizing Mediterranean elements, coupled with weekly group sessions (10 weeks; ~50 hours of nutrition education including lectures and cooking classes, plus lifestyle counseling). DASH-only arm received the same education program without the fasting pre-phase. Primary outcomes: 24 h ambulatory systolic blood pressure (SBP) at week 12 and HOMA-index at week 12. Clinical measurements: 24h ambulatory blood pressure monitoring (ABPM; Mobil-O-Graph PWA), office BP, body weight, BMI, waist/hip circumferences, body fat percentage (bioimpedance), and laboratory parameters (glucose, insulin, HbA1c, lipids, creatinine, eGFR, CRP, IL-6, etc.). Medication usage was recorded and normalized to track dose changes. Immunophenotyping: Peripheral blood mononuclear cells were isolated and profiled via multicolor flow cytometry (FACSCanto II). Panels included major leukocyte lineages and activated/functional T-cell subsets (e.g., Th1/Th17, MAIT cells, Tregs, CD8+ subsets). FlowSOM was used for unbiased clustering. Cytokine production assessed after PMA/ionomycin restimulation. Microbiome profiling: Stool samples underwent 16S rRNA gene sequencing (V4 region) and shotgun metagenomics. Reads were processed with LotuS (16S) and NGLess (shotgun), mapped to SILVA v138, mOTUv2, and the IGC gene catalog. Functional profiles binned to KEGG KOs/modules and Gut Microbial Modules (GMM). Rarefaction employed RTK. Enterotypes assessed with Dirichlet Multinomial mixtures. Statistics: Medication changes, age, and sex were controlled in longitudinal models. Community composition changes evaluated via Bray-Curtis (microbiome) or Euclidean distances (immunome) with PERMANOVA (adonis). Univariate longitudinal tests used nested linear model comparisons with BH FDR correction (FDR<0.1 for selection, then post-hoc Mann-Whitney U with FDR<0.05). Effect sizes reported as Cliff’s delta. Correlation/network analyses used Spearman correlations (FDR<0.05) with mixed-effects confirmation accounting for repeated measures. Missing data handled via multiple imputation (MCMC). Machine learning: Logistic regression with L2 penalty; leave-one-subject-out cross-validation; forward stepwise selection to identify top 10 features for prediction. External validation used a published fasting cohort (Mesnage et al.) reprocessed through the same pipeline. Safety: Adverse events recorded; common fasting symptoms were mild (weakness, headache, hunger); no serious adverse effects reported.

- Microbiome and immunome shifts with fasting: Fasting induced substantial, significant compositional shifts in the gut microbiome (PERMANOVA P=0.001) and immunome (PERMANOVA P=0.001) during the fasting week (V1→V2), largely reversing after refeeding (V2→V3). Alpha and beta diversity changes were not significant. - Taxonomic and functional changes: During fasting, many Clostridial Firmicutes, including key butyrate producers (Faecalibacterium prausnitzii, Eubacterium rectale, Coprococcus comes), decreased and were restored after refeeding. Bacteroidaceae showed the opposite trend. At 3 months, Enterobacteriaceae, particularly Escherichia coli, were persistently depleted. Functional capacity shifted markedly, with enrichment of propionate production modules (MF0121, MF0126), mucin degradation, and diverse nutrient utilization pathways. - Clinical outcomes: Compared to DASH alone, fasting + modified DASH significantly reduced 24h ambulatory SBP and MAP at 3 months (two-sided MWU, FDR<0.05). DASH alone did not significantly change 24h SBP (P=0.27). Office SBP decreased under DASH at 3 months, but gold-standard 24h ABPM improvements were significant only with fasting + DASH. Antihypertensive medication reduction occurred in 43% of fasting participants vs 17% on DASH (χ² P=0.035), with BP remaining controlled. BMI and body weight decreased significantly and durably only in the fasting arm (P<0.001), and these changes did not explain the BP improvements or omics shifts (95% of significant findings remained after adjusting for BMI change). - Responder analysis: In the fasting arm, responders (n=22) had median SBP decrease of 8.0 mmHg vs 0.3 mmHg in non-responders (n=10). DASH responders (n=17) also had 8.0 mmHg median reduction, but many fasting-related microbiome/immunome effects did not replicate in DASH-only. Responder-specific microbiome shifts included sustained enrichment of an unclassified Clostridium sp., and refeeding-associated enrichment of F. prausnitzii, alongside depletion of Actinomycetaceae/Actinomyces and Sphingomonas. Functionally, responders showed stronger enrichment of propionate production modules and unique changes (e.g., pyruvate:formate lyase MF0085 decreased during recovery only in responders). Immune changes (e.g., decreases in Th17/Th1 cytokine-producing cells) were more pronounced in responders. - Cross-system correlations: Circulating cytokine-producing MAIT cells (IL-2+, TNFα+, IFNγ+) positively correlated with 24h ambulatory SBP (Spearman’s rho ~0.44–0.50; FDR≤0.044). Non-classical monocytes positively correlated with MAP (rho=0.73; FDR=0.002). Several SCFA-producing taxa (e.g., E. rectale, Dorea longicatena) negatively correlated with pro-inflammatory CD4+ cytokine producers and MAIT subsets. Regulatory T cell-like subsets correlated with BMI/weight. - Predictive modeling: At baseline, immunome-based logistic regression predicted fasting BP response with 71% accuracy (sensitivity 75%, specificity 70%, F1 77%); using longitudinal immune changes increased accuracy to 78%. Key immunologic predictors indicated a less pro-inflammatory baseline signature in responders (e.g., lower Th1/Treg ratio, fewer IL-17+TNFα+ MAIT, fewer CD24+ memory CD8+ cells). Microbiome-based features at baseline also characterized responders (e.g., depletion of Desulfovibrionaceae and propionate biosynthesis genes), which increased during fasting/refeeding. - External validation: A microbiome-based model trained on baseline responder-specific taxa predicted BP response in an independent fasting cohort (Mesnage et al., healthy men) with ~67% accuracy (correctly classifying 10/15). Top contributors included Desulfovibrionaceae, Hydrogenoanaerobacterium, Akkermansia, and Ruminococcaceae GCA-900066225. - Safety: Only mild, transient fasting-related symptoms were reported; no serious adverse events.

The study demonstrates that a brief, structured fasting period before initiating a plant-forward DASH diet leads to coordinated shifts in the gut microbiome and immune system and yields sustained improvements in 24h ambulatory SBP and MAP in hypertensive MetS patients. The findings support a preconditioning hypothesis: fasting transiently resets microbiome and immune states, enabling subsequent DASH refeeding to promote enrichment of SCFA-producing taxa and functional modules (notably propionate pathways), which may reduce vascular inflammation and improve hemodynamics. Importantly, BP improvements were not explained by weight loss alone, and most omic effects persisted after adjusting for BMI change. Cross-system network analyses linked BP to specific immune populations (MAIT cells, non-classical monocytes) and SCFA-producing microbes, suggesting a mechanistic axis where increased SCFA availability modulates immune tone and vascular function. Heterogeneous BP responses could be anticipated by baseline immunome/microbiome features, indicating potential for precision nutrition: individuals with depleted SCFA capacity (e.g., Desulfovibrionaceae lower; propionate modules reduced) benefitted more, and transient restoration during fasting/refeeding corresponded to sustained BP control. External validation of microbiome predictors in an independent fasting cohort, and parallels with metformin-associated functional signatures, support generalizability of core mechanisms.

A 5-day fasting intervention followed by a modified DASH diet in hypertensive MetS patients induces specific, largely reversible microbiome and immunome changes, while conferring sustained reductions in ambulatory SBP/MAP, body weight, and antihypertensive medication needs. Multi-omics integration and machine learning identified immune and microbial signatures associated with, and predictive of, durable BP response, emphasizing SCFA-related pathways and MAIT cell dynamics. These results position periodic fasting as a promising, non-pharmacological adjunct to dietary therapy in MetS-related hypertension. Future research should (1) validate predictive immunome and microbiome biomarkers in larger, diverse cohorts; (2) experimentally interrogate causal pathways (e.g., SCFAs, MAIT cell modulation) in animal models; (3) define optimal fasting frequency/duration and the necessity/format of refeeding diets; and (4) evaluate long-term cardiovascular outcomes and scalability in clinical practice.

- Generalizability: Cohort consisted of Caucasian-European participants, introducing selection bias; applicability to more diverse populations is uncertain. - Sample size: Modest number of participants reduces power to detect subtler effects and limits completeness of identified features. - Blinding: Participants could not be blinded to intervention; although laboratory personnel were blinded, expectancy and behavioral differences may influence outcomes. - Intervention components: Study design does not isolate the long-term effects of fasting without subsequent DASH; synergy is likely, but precise contributions of each component cannot be disentangled here. - Frequency/durability: Optimal repetition schedules for fasting cycles and long-term durability beyond 3 months were not assessed. - Causality: Observational correlations and network analyses cannot establish causal mechanisms; experimental validation (e.g., gnotobiotic models, SCFA interventions) is needed. - External validation: Immunome-based predictive models lack validation in independent fasting cohorts; only microbiome predictors were externally tested. - Medication adjustments: Although statistically controlled, medication changes could still confound some clinical and omic associations.