Longevity & AgingResearch PaperOpen Access

Scientists Map 661 Gut Microbes Linked to Heart and Metabolic Health in 34,000 People

A landmark study across US and UK populations identifies gut microbiome species consistently tied to cardiometabolic risk, diet quality, and disease markers.

Friday, July 3, 2026 1 view
Published in Nature
Colourful 3D gut microbiome cross-section with diverse microbial species floating in a luminous intestinal landscape alongside fresh vegetables and grains

Summary

Researchers analyzed gut microbiome data from over 34,000 US and UK adults alongside detailed dietary, anthropometric, and clinical health markers. Using metagenomics and machine learning, they identified 661 microbial species and ranked them by their association with cardiometabolic health outcomes. The resulting 'ZOE Microbiome Health Ranking 2025' distinguished favourably and unfavourably associated species reproducibly across five independent cohorts. The ranking was validated in over 7,800 public samples and two dietary intervention trials, where healthier-ranked species increased after dietary improvements. Most top-ranked species belong to the Firmicutes phylum, particularly Clostridia. The study underscores the gut microbiome's role as a modifiable link between diet and metabolic disease, while cautioning that causal conclusions require prospective and interventional studies.

0:00--:--

Detailed Summary

Cardiometabolic diseases, including cardiovascular disease and type 2 diabetes, are the leading causes of death in Western countries, and both poor diet and gut microbiome composition have been implicated in their development. Yet large-scale, multi-national studies systematically linking specific microbial species to health markers across diverse populations have been lacking—until now.

This study assembled and harmonized five of the largest gut metagenomic cohorts to date from the ZOE PREDICT programme, encompassing 34,694 participants from the United States and United Kingdom. Each participant provided stool samples for shotgun metagenomic sequencing alongside detailed dietary records, anthropometric measurements (including BMI), and a comprehensive panel of 37 clinical markers spanning glycemia, lipid profiles, inflammatory markers such as GlycA, blood pressure, and atherosclerotic cardiovascular disease (ASCVD) risk scores.

Using a systematic machine learning approach—including random forest classifiers and regression models trained on species-level relative abundances—the researchers found consistent, moderately strong associations between the microbiome and surrogate health markers across all five cohorts. Key predicted markers included fasting and postprandial glycemia, triglycerides, cholesterol, and inflammatory indices, with AUC values ranging from 0.64 to 0.73 and Spearman correlations of 0.30–0.46. Dietary quality indices such as the Healthy Eating Index and the Healthful Plant-Based Diet Index also correlated meaningfully with microbiome composition.

To distil actionable insight, the team computed partial Spearman's correlations (adjusted for sex, age, and BMI) between each of 661 prevalent microbial species and all 37 health markers, then averaged and ranked species across cohorts. This produced the 'ZOE Microbiome Health Ranking 2025,' a reproducible summary ranking microbes from most favourably to most unfavourably associated with host health. The majority of top-ranked species—both favourably and unfavourably—belonged to the Firmicutes phylum and Clostridia class, particularly the Lachnospiraceae family, highlighting that beneficial and harmful microbes can coexist within the same taxonomic group. The ranking was independently validated in over 7,800 publicly available metagenomic samples and showed strong associations with BMI and disease conditions in those external datasets.

Critically, in two separate dietary intervention clinical trials totalling 746 participants, adherence to healthier dietary patterns led to increases in the abundance and prevalence of favourably ranked species and reductions in unfavourably ranked ones—providing preliminary interventional support for the ranking's biological relevance. The authors emphasise, however, that these are observational associations and that establishing causality requires prospective cohort studies and rigorously designed interventional trials. The ZOE Microbiome Health Ranking 2025 is nonetheless positioned as a principled, large-scale foundation for guiding future mechanistic and causal research into gut microbiome-targeted dietary interventions for cardiometabolic health.

Key Findings

  • A microbiome health ranking of 661 gut species was derived from 34,694 US and UK adults using 37 cardiometabolic markers.
  • Machine learning models linked gut microbiome composition to glycemia, triglycerides, cholesterol, and inflammation with AUCs of 0.64–0.73.
  • The ZOE Microbiome Health Ranking 2025 was reproducible across 5 independent cohorts and validated in 7,800+ public samples.
  • In two dietary intervention trials (n=746), healthier-ranked microbes increased and unfavourably ranked microbes decreased with improved diet.
  • 92% of the top 50 ranked species (both favourable and unfavourable) belonged to Firmicutes, mainly the Clostridia class.

Methodology

Five cross-sectional ZOE PREDICT cohorts (n=34,694, US and UK) were analysed using shotgun metagenomics at the species-level genome bin (SGB) level. Random forest machine learning models and partial Spearman's correlations (adjusted for sex, age, BMI) linked 661 microbial species to 37 clinical markers. Validation used 7,800+ public metagenomic samples and two independent dietary intervention RCTs (n=746).

Study Limitations

The study is observational and cross-sectional for the primary cohorts, precluding causal inference about whether microbiome changes drive or merely reflect health outcomes. Participants in PREDICT cohorts were largely healthy adults from the US and UK, limiting generalisability to other ethnicities, disease populations, or dietary cultures. The intervention trials, while supportive, were not powered to disentangle microbiome-mediated effects on health from direct dietary effects.

Enjoyed this summary?

Get the latest longevity research delivered to your inbox every week.

Enter your email to subscribe: