CAG-170: the hidden gut bacteria found in healthy people worldwide.

The study, properly framed

Cambridge University announced in May 2026 the publication of a large-scale meta-analysis in Cell Host & Microbe [da Silva 2026]. The team — led by researchers at the Wellcome Sanger Institute and EMBL-EBI — pulled 11,115 gut metagenomes from 39 countries spanning every inhabited continent. They standardised processing across 13 publicly available cohorts, layered participant phenotype data (disease status, demographics, diet pattern where available), and asked a simple question with unprecedented statistical power: which microbes are consistently higher in healthy people across populations, and consistently lower in people with chronic disease?

The answer, surprisingly, did not lead with any of the famous names. Not Akkermansia. Not Faecalibacterium prausnitzii. Not Bifidobacterium. The strongest single health-associated signal across the entire dataset was an uncultured bacterial lineage the authors call CAG-170 — a genus-level group that has been showing up in metagenomes for years but has never been successfully grown in a laboratory and therefore lacks the proper Linnaean name that a culturable bacterium would have.

CAG-170 was higher in healthy people across every continent in the dataset. It was depressed in inflammatory bowel disease (IBD), in colorectal cancer, in type 2 diabetes, in cardiovascular disease, and across multiple liver conditions. The association held after adjustment for age, geography, and diet pattern. It was, in the authors' words, the most robust cross-cohort signature of "general gut health" the analysis produced.

What CAG-170 actually is (and isn't)

The naming convention matters because it signals what we know and don't know. CAG stands for "Co-Abundance Group" — a grouping framework used in metagenomic work where you can see a coherent cluster of microbial DNA that consistently rises and falls together across samples, but you have not been able to isolate the organism in pure culture to formally describe it. CAG-170 is the 170th such group catalogued in the reference framework the field uses.

What that means in plain terms: we can read CAG-170's genome out of stool samples. We know roughly what its biology should be from its gene content. We do not have a CAG-170 sitting in a Petri dish anywhere in the world. We have not watched it grow, we have not tested what conditions it needs, we have not characterised its metabolism in vitro. Every CAG-170-specific intervention you might imagine — a stool-test marker, a probiotic capsule, a screen for gut health — is downstream of culturing it, and the culturing problem remains unsolved.

The Cambridge press release was upfront about this. CAG-170 is one of the strongest health-associated bacteria identified in the microbiome era, and we know it primarily as a genetic signature. That gap between "we know it matters" and "we can do anything directly about it" is the single most important thing to hold in mind when reading the rest of the press coverage.

Why an uncultured bacterium tops the list

The fact that the strongest signature of gut health is something nobody has cultured tells you something structural about the microbiome field. Most of the bacteria the public knows the names of (Lactobacillus, Bifidobacterium, Akkermansia, Faecalibacterium) are exactly the ones that can be cultured — which is why they show up in probiotics, in stool-test panels, and in headlines. Cultureability is partly a feature of the bacterium and partly a feature of which laboratories have invested in the right media. The uncultured majority of the human gut microbiome has been understudied not because it is less important but because it is harder to handle.

The 2026 meta-analysis is part of a broader correction. As shotgun-metagenomic sequencing has become cheap enough to run on thousands of samples, the dominant view of which microbes drive gut health has shifted from "the ones we can put in a capsule" toward "the ones the data says correlate with health, whether or not we can grow them." CAG-170 is the most visible example so far.

It is not the only one. The same paper identified additional uncultured lineages that track health-disease gradients across the dataset. CAG-170 leads the list because the effect size was largest and most consistent, but the broader picture is that a substantial portion of the "healthy gut" signal lives in microbes the supplement industry literally cannot sell you.

The strongest signal of gut health across 11,115 people from 39 countries is a microbe we cannot grow, cannot supplement, and only know from its genetic shadow.

What it appears to do — B12, fibre, and the unknowns

From the genome reconstructions, the Cambridge team could read the functional capacity CAG-170 carries even without culturing it. Two functions stood out as candidate mechanisms.

First, CAG-170 carries the full enzymatic pathway for vitamin B12 biosynthesis. B12 is one of a small number of vitamins humans cannot make and must obtain from diet or from microbial synthesis in the gut. Most gut bacteria cannot make B12 — the genes for the full pathway are scarce in the microbiome. That CAG-170 carries the complete pathway suggests it may be a meaningful source of B12 or B12-related cofactors for the rest of the gut community, which in turn affects host vitamin status.

Second, the genome reconstruction is consistent with a complex-fibre cross-feeding role. CAG-170 appears tightly integrated into the carbohydrate-handling networks of the gut — receiving fermentation intermediates from upstream fibre-degraders and contributing its B12 output downstream. That ecological position is consistent with CAG-170 sitting in the fibre-handling backbone of a healthy gut, even if it is not the primary fibre-degrader itself.

The authors are explicit about what CAG-170 is not able to do: the reconstructed genome lacks key genes involved in arginine biosynthesis, and several other amino-acid pathways are partial. The picture isn't of a self-sufficient organism but of one tightly integrated into cross-feeding networks — it depends on the rest of a diverse community to thrive, which is part of why it disappears when the community thins out.

These are inferences from genomic content. The Cambridge group is clear that they have not directly demonstrated B12 production in vivo or in vitro from CAG-170 (you cannot, without culturing it), and the carbohydrate enzymes are mapped from sequence homology rather than measured activity. The functions are plausible, consistent with the cross-population health association, and worth following up. They are not yet proven.

Why there is no supplement, and won't be soon

Press coverage of any microbiome paper that uses words like "healthy" and "found worldwide" gets a predictable next-day wave of supplement marketing. With CAG-170, that wave will be either absent or deceptive, and it is worth being clear why.

You cannot package a microbe you cannot grow. Probiotic manufacturing depends on producing large quantities of viable bacteria in fermentation tanks, freeze-drying them, encapsulating them, and shipping them at quantities and viability levels the label claims. Every step depends on knowing how to culture the organism. CAG-170 has not been cultured. Until somebody works out the culture conditions — and that work has been actively attempted for years without success on similar lineages — no legitimate CAG-170 probiotic can exist.

Anything that markets itself as a CAG-170 probiotic in the near-term is one of three things: a different organism the marketer renamed, a fermentation by-product without live cells, or fraud. None of those will deliver the bacterium the Cambridge team flagged.

Stool-microbiome tests can in principle measure CAG-170 abundance from sequencing data — that is how the paper measured it. Whether commercial direct-to-consumer microbiome tests will add CAG-170 as a reported metric is a question of how fast their bioinformatics pipelines incorporate the new reference. For most readers, the signal is not actionable enough yet to justify the test cost.

What this means for what you do now

The honest practical version. The supplement layer is closed off for now. The substrate layer — what you feed your gut community — remains the only intervention CAG-170 connects to.

CAG-170's carbohydrate-active-enzyme profile suggests it is a complex-fibre specialist. The food intake that supports complex-fibre specialists is exactly the food intake that supports the rest of the healthy-gut bacterial roster identified in this and prior cross-cohort work: diverse plant fibre, with an emphasis on the harder-to-digest end of the spectrum (resistant starch, whole grains, legumes, cruciferous vegetables, root vegetables eaten in skins). The general direction is consistent with what the rest of the gut-health literature has been saying for a decade. CAG-170 does not change the prescription; it adds one more reason the prescription is the right one.

On the B12 layer, there is a small inferential implication. If CAG-170 is a meaningful microbial B12 source, then chronically low CAG-170 (which tracks Western-pattern diets and chronic disease) may be one input into the population-level B12 status problems documented in older adults and in long-term vegetarians and vegans. It does not change the standing advice to supplement B12 if you are vegan, vegetarian, on metformin, on PPIs, or over 60 — but it does add a structural reason to take the gut substrate seriously even while supplementing.

On the cross-disease pattern: CAG-170 is depressed across IBD, T2D, cardiovascular disease, and colorectal cancer in this dataset. That does not mean CAG-170 is causal for any of those conditions. It does mean that the kind of fibre-rich, low-ultra-processed, low-antibiotic-burden lifestyle that supports CAG-170 is the same lifestyle the rest of the metabolic and inflammatory literature has been pointing to. The signals keep converging on the same substrate work.

A tiered framework

Frameworks, not prescriptions. With that established: