Tuesday, September 29, 2026

Two Months or Ten Years?

How much new cancer drugs really extend life — a fact-check of the "2.1 months" claim, updated with the evidence since 2014

A number keeps circulating in discussions about cancer treatment: new cancer drugs extend life by an average of just 2.1 months. It appeared again this September in a NutritionFacts.org post arguing that most new chemotherapy drugs deliver trivial benefits, that patients are rarely told this, and that drugs stay on the market even after they fail.

Much of that is correct. But the 2.1-month figure is also one of the most misunderstood numbers in oncology. It is a median across dozens of drugs, not a promise for any patient, and it hides the fact that cancer drug benefits are not spread evenly. Most approvals add a little. A few add years, for specific people.

The typical new cancer drug still adds only a couple of months. The exceptional ones add years — but only for a defined subset of patients. Both statements are true at once.

            A note on scope. This is an evidence review, not treatment advice. Survival statistics describe groups, not individuals. Anyone weighing a treatment should discuss the expected benefit for their specific cancer, stage, and biomarkers with their oncologist.
        

Part I / What the Critics Get Right

The 10-Day Survival Gain

The classic example is erlotinib added to gemcitabine for advanced pancreatic cancer. In the pivotal PA.3 trial, median overall survival was 6.24 months versus 5.91 months. The result was statistically significant, and the drug was approved. The difference in medians is about 10 days.

Two details are usually left out. The comparison group was not on placebo alone — they received gemcitabine chemotherapy plus a placebo. And the median undersells the effect slightly: one-year survival was 23% versus 17%. Still, the benefit was genuinely small, and it came with added side effects and cost.

statistically significant ≠ clinically meaningful ≠ what patients hear as "longer survival"

The 2.1-Month Median

The headline figure comes from Fojo and colleagues (2014), who examined 71 drugs approved by the FDA for solid tumors between 2002 and 2014. The median gain was 2.5 months in progression-free survival and 2.1 months in overall survival, for drugs that typically cost over $100,000 per year.

Other findings in the same critical literature also hold up:

Failed drugs stay on the market. Rupp & Zuckerman (2017) looked at 18 cancer drugs approved 2008–2012 that did not later show an overall survival benefit. Most kept FDA approval at premium prices. The most expensive, cabozantinib for thyroid cancer, actually worsened patients' symptom scores.
Patients misunderstand the goal of treatment. In Weeks et al. (NEJM 2012), 69% of patients with metastatic lung cancer and 81% with metastatic colorectal cancer did not report understanding that their chemotherapy was very unlikely to cure them.
Preferences vary enormously. In Silvestri et al. (BMJ 1998), some lung cancer patients would accept chemotherapy for one extra week; others would refuse it even for two years. Only 22% would choose chemo for a 3-month survival gain — but 68% would choose it if it substantially relieved symptoms without extending life.

That last point is often dropped. For many patients, the value of treatment is not measured only in months.

Part II / Where the Framing Breaks

Fact-Checking the Claims

Checked against the original studies, most individual numbers in the critique are accurate. The problems are in how they are framed.

Claim	Verdict	What the source actually says
Erlotinib added ~10 days in pancreatic cancer	Accurate	6.24 vs 5.91 months median; control arm received gemcitabine, not placebo alone; 1-year survival 23% vs 17%.
New chemotherapy drugs average 2.1 months	Misleading	The 71 drugs were mostly targeted therapies and biologics, not classic chemo; limited to solid tumors; 2.1 months is a median, not an average.
Ineffective drugs keep approval at high prices	Accurate	Matches Rupp & Zuckerman (2017), though the FDA published a rebuttal disputing parts of that analysis.
~Three-quarters of patients misunderstand cure	Accurate	69% (lung) and 81% (colorectal) in patients with metastatic disease.
Chemo adds only 1–2% to 5-year survival in common cancers	Contested	From Morgan et al. (2004): 2.3% (Australia) and 2.1% (USA) overall. Critics argued the method dilutes the benefit by counting all newly diagnosed patients, including those who never needed chemo. Data are over 20 years old.
This describes cancer drugs today	Outdated	A post dated 2026 relying on 1990–2019 evidence omits immunotherapy and targeted-therapy results that changed outcomes in several cancers.

Individual numbers mostly hold. The framing — "chemo," "average," and "today" — does not.

Part III / The Newer Evidence

The Typical Benefit Hasn't Moved

If the 2.1-month figure were an artifact of one era, newer analyses would show it rising. They don't. Every large review that extends the window lands in the same narrow band.

Analysis	Period	Median OS gain
Fojo et al. — 71 FDA solid-tumor approvals	2002–2014	2.1 months
Ladanie et al. — 92 novel FDA-approved drugs	2000–2016	2.40 months
FDA evidence base — 145 novel drugs, 156 indications	2000–2020	2.55 months
Swissmedic approvals — meta-analysis	2001–2020	2.42 months

Two decades of approvals, four independent analyses, one consistent answer: roughly 2–2.5 months.

The newer reviews add detail that makes the picture worse, not better. Of 31 drugs approved 2000–2016 with median survival data, only one improved survival by more than six months. Half of all novel cancer indications through 2020 were approved without randomized trial evidence. And outside trials, results tend to shrink: when trial outcomes for 22 drugs were compared with real-world data from older Medicare patients, survival was shorter in 28 of 29 indications, by a median of 6.3 months.

Accelerated approvals remain a weak point. According to a 2025 JAMA analysis, most cancer indications granted accelerated approval between 2013 and 2017 still had not shown a benefit in survival or quality of life by mid-2023.

What the Median Cannot See

A median survival gain measures the midpoint of a survival curve. It is blind to the tail — the patients who are still alive years later. For most drugs there is no meaningful tail. For some, the tail is the whole story.

Immune checkpoint inhibitors in advanced melanoma are the clearest case. In the CheckMate 067 trial, median overall survival was 71.9 months with nivolumab plus ipilimumab, versus 19.9 months with ipilimumab alone. At ten years, 43% of patients on the combination were alive. A generation earlier, metastatic melanoma was almost uniformly fatal within a year or two.

The Typical Approval

MARGINAL, BROAD

Median OS gain of ~2–2.5 months
Often approved on surrogate endpoints
Benefit shrinks in real-world patients
Survival curves converge; no durable tail
Priced like breakthroughs

The Transformative Minority

LARGE, NARROW

Median gains measured in years
A durable plateau of long-term survivors
Visible in national mortality statistics
Concentrated in biomarker-defined groups
Melanoma, driver-mutation lung cancer, some blood cancers

Even among drugs approved before survival data matured, the split is sharp. Of 38 such indications, only 12 (32%) later proved a survival benefit — but where they did, the median gain was 12.5 months. The average of a large, near-zero group and a small, large-effect group produces a modest number that describes neither.

Part IV / Who Actually Benefits

Broad Eligibility, Narrow Response

The obvious follow-up question: how many patients fall into the lucky subset? Researchers who are openly skeptical of oncology hype — Alyson Haslam and Vinay Prasad — have tracked this for years. Their numbers are best-case estimates: they assume every eligible patient receives the drug and use response rates from drug labels.

Checkpoint inhibitors · 2023

56.6% → 20.1%

Share of US patients with advanced cancer eligible for immunotherapy, versus the share estimated to respond. Up from 1.5% and 0.1% in 2011. Haslam et al., 2025

Genome-targeted drugs · 2020

13.6% → 7.0%

Share eligible for a drug matched to a tumor mutation, versus the share estimated to respond. Up from 5.1% and 2.7% in 2006. Haslam et al., 2021

Sources: Haslam, Olivier & Prasad (2025); Haslam, Kim & Prasad (2021). Response is not the same as long-term survival, so the share gaining years of life is smaller still.

So the common perception — that effective new drugs help only a small subset — needs one refinement. Immunotherapy is no longer niche in who receives it; more than half of patients with advanced cancer are now eligible. But roughly one in five responds, and a smaller fraction of those achieves the durable, melanoma-style survival that makes headlines. Targeted drugs help a much smaller group, sometimes dramatically.

When the Tail Shows Up in National Statistics

Where benefits are concentrated and real, they eventually become visible at the population level — something the older critiques could not show. A 2020 NEJM analysis by the US National Cancer Institute found that mortality from non-small cell lung cancer fell faster than its incidence, with the decline accelerating in 2013 — right after routine testing for EGFR and ALK mutations and matched targeted therapy became standard. The authors estimated that nearly 10,000 deaths were delayed between 2014 and 2016.

            The natural control group. Small cell lung cancer, which has no comparable targeted therapies, also saw falling deaths over the same period — but entirely because fewer people developed it. Survival after diagnosis did not improve. Same organ, same era, same health system; the difference was whether an effective drug existed for the tumor's biology.
        

biomarker test → matched drug → large benefit in a subgroup → visible drop in national mortality

Part V / What It Means

Two Truths at Once

The critique of marginal cancer drugs survives the newer data intact. Twenty years of approvals keep producing the same ~2-month median, regulators continue to approve drugs on surrogate endpoints, failed drugs linger, and prices bear little relation to benefit. Patients deserve to hear those numbers plainly.

But the broader implication — that modern cancer drugs rarely change outcomes — is outdated. The real pattern is lopsided: a large majority of marginal approvals and a small minority of transformative ones, concentrated in specific tumors and biomarker-defined groups. Averaging the two produces a figure that misrepresents both.

            Questions worth asking about any new cancer drug:
            What was it compared against? Placebo, older chemo, or an active standard of care?
Overall survival or a surrogate? Tumor shrinkage and progression-free survival do not always translate into longer life.
Median or tail? What fraction of patients were alive at 3 or 5 years in each arm — is there a plateau?
Does my tumor match the trial population? Stage, prior treatments, and biomarkers (EGFR, ALK, PD-L1, MSI, etc.) often decide which group you belong to.
What does it do to quality of life? Symptom relief can matter as much as time.

        

The honest answer to "how much do new cancer drugs extend life?" is not two months, and it is not ten years. It is: it depends almost entirely on which drug, for which tumor, in which patient — and the most useful thing a patient can ask is which of those two groups their treatment belongs to.

Tuesday, September 8, 2026

MYC and Ammonia

Cancer metabolism / Research perspective

MYC and Ammonia: When Cancer’s Waste Talks Back

A possible feedback loop connects an oncogene, nitrogen metabolism, and immune suppression. Could interrupting it expose a new cancer vulnerability?

Imagine a cancer cell changing its surroundings as it grows. The nutrients it consumes leave chemical traces. Some become building materials for neighboring cells. Others interfere with the immune cells trying to destroy the tumor. A molecule that begins as metabolic waste can become part of the environment that helps the cancer persist.

Ammonia brings this possibility into focus. Its connection to MYC, a protein that regulates growth and metabolism, raises a provocative question: does MYC create the ammonia-rich environment, or can ammonia help sustain MYC? The answer may involve both directions—but the evidence for them is unequal.

A tumor’s metabolic output may help maintain the conditions that support it. If ammonia participates in that maintenance, clearing it could matter even when MYC itself remains difficult to inhibit.

This article examines experimental findings and research proposals. The ammonia-targeting approaches discussed here have not established clinical benefit for treating cancer in people.

Part I / The nitrogen balance

How MYC Can Increase Ammonia Production

MYC helps coordinate the supply of nutrients needed for growth. In a foundational study, reducing MYC in a human glioma cell line lowered glutamine consumption and ammonia production. This provides experimental evidence for a forward connection: MYC can drive metabolism that releases ammonia. [Wise et al., 2008]

One route runs through glutaminase, or GLS, which converts glutamine into glutamate and releases ammonia. Researchers subsequently showed that MYC can increase GLS expression by repressing microRNAs that normally restrain it. [Gao et al., 2009]

MYC activity → increased glutamine use → ammonia production

Production, however, is only one side of the balance. Local ammonia also depends on reassimilation, disposal, and movement between cells and tissues. A tumor that produces ammonia rapidly will not necessarily accumulate it if removal keeps pace.

The urea-cycle connection needs qualification. Impaired ammonia disposal can contribute to accumulation, as demonstrated in colorectal cancer research involving HNF4α and the urea-cycle enzyme OTC. That finding does not establish a universal sequence in which MYC switches off the urea cycle. [Bell et al., 2023]

MYC can also increase glutamine synthetase, or GS, which captures ammonia and combines it with glutamate to make glutamine. In the studied models, this supported survival when glutamine was scarce and supplied nucleotide synthesis. MYC therefore has the capacity to promote both ammonia-producing and ammonia-consuming pathways. Their relative activity matters more than the MYC label alone. [Bott et al., 2015]

Throughout this article, “ammonia” refers broadly to the interconverting NH₃/NH₄⁺ pool. Their relative proportions depend on pH; the two forms should be distinguished when interpreting experiments.

Waste That Can Be Reused

Recovering Nitrogen

Isotope-tracing experiments showed that breast cancer cells can incorporate ammonia nitrogen into glutamate through glutamate dehydrogenase, then transfer it into other amino acids. This recycling supported tumor biomass. Describing ammonia as being “turned into DNA” skips the metabolic intermediates that make nitrogen usable. [Spinelli et al., 2017]

Adjusting Cell Recycling

Ammonia released by glutamine metabolism can regulate autophagy, the process through which cells recycle internal components. In experimental systems, this response helped cells tolerate stress. Its effect depends on exposure and cellular context; ammonia is not an unlimited growth resource. [Eng et al., 2010]

Part II / The immune cost

An Environment That Weakens Tumor Killing

A metabolite useful to a cancer cell can be harmful to an immune cell beside it. Colorectal cancer experiments linked ammonia exposure to impaired T-cell proliferation and function, with increased exhaustion-associated features. This gives ammonia a possible role in maintaining immune suppression. [Bell et al., 2023]

A separate study identified a more specific failure in the killing machinery. Ammonia increased pH inside acidic cellular compartments and reduced mature perforin, a protein required for effective cytotoxic attack. NK cells and engineered CAR-T cells consequently killed cancer cells less effectively in vitro. [Domagala et al., 2025]

Ammonia exposure → less acidic secretory compartments → less mature perforin → weaker killing

The immune cell may still recognize and contact its target while delivering a less effective attack. These findings also concern pH inside cellular compartments; they do not show that the whole tumor becomes alkaline. [Domagala et al., 2025]

Different immune cells can respond differently. A liver cancer study published in Cell in 2026 found that regulatory T cells could adapt to ammonia and strengthen their suppressive function. Together, these observations suggest a local imbalance: reduced activity of tumor-killing cells alongside support for cells that restrain immunity. [Gu et al., 2026]

Part III / The chicken-and-egg question

Can Ammonia Send a Signal Back to MYC?

There is an experimental clue. In a 2025 study using Huh7 liver cancer cells, a lower ammonium chloride exposure increased c-MYC protein and active β-catenin, alongside an autophagy response. Higher exposures produced a declining MYC response. Manipulating MYC supported its involvement in the autophagy effect. [Sergio et al., 2025]

That is evidence for a possible return signal in an established cancer cell line. It does not show that ammonia initiates cancer, directly stabilizes MYC protein, or causes enough additional ammonia production to close the loop. Nor does it establish β-catenin as a necessary intermediate. The concentration dependence also argues against assuming that more ammonia always means more MYC.

Initiation and maintenance are different questions. A genetic change could activate MYC first, while a later metabolic consequence helps sustain the tumor. Alternatively, impaired clearance could precede elevated ammonia. Determining what keeps the system running may reveal an intervention even before its original starting point is known.

Connection	What the evidence supports	What remains open
MYC → ammonia	MYC-dependent glutamine metabolism can increase production.	Whether local production exceeds removal in a particular tumor.
Ammonia → MYC	A response observed in a limited experimental context.	Its generality and whether it completes a sustained feedback loop.
Ammonia → immune suppression	Functional impairment demonstrated in several experimental systems.	Which patients would benefit from an ammonia-directed intervention.

These are separate evidence streams. Combining their arrows produces a research hypothesis, not proof of the complete circuit.

Part IV / Where to intervene

Targeting the Environment MYC Helps Create

Direct MYC targeting remains an active research field: the investigational inhibitor OMO-103 reached a phase I trial that demonstrated target engagement. The ammonia perspective adds another possibility—intervening in a metabolic consequence of MYC activity. [Garralda et al., 2024]

Three approaches deserve comparison. The practical question is which one improves the balance between cancer survival and immune attack.

Research direction 01

Reduce Tumor Production

Test whether suppressing tumor MYC activity or its glutaminase-dependent metabolic output lowers local ammonia. MYC perturbation provides a biological rationale, but it cannot predict the result in every cancer. Measure ammonia alongside immune-cell function; a pathway change alone does not establish an immune benefit. [Wise et al., 2008]

Research direction 02

Increase Local Clearance

In colorectal cancer mouse models, ammonia clearance improved immune activity, and ornithine combined with anti-PD-L1 improved survival. This supports testing clearance as an immunotherapy partner. It does not establish that systemic ammonia-lowering drugs or supplements reproduce that effect in people. [Bell et al., 2023]

Research direction 03

Protect the Immune Cells

Memory CD8 T cells can use urea and citrulline cycle pathways to dispose of ammonia. This suggests an engineering question: could therapeutic immune cells be equipped to retain function in ammonia-rich tumors? Enhancing resilience would need to preserve killing capacity as well as survival. [Tang et al., 2023]

An unconventional extension

Convert a Local Waste Stream

Engineered bacteria have been designed to convert tumor ammonia into arginine, supporting antitumor immunity and combining with checkpoint blockade in mice. This offers a local metabolic intervention. Because arginine itself benefits T cells, the antitumor effect cannot be attributed solely to ammonia removal. [Canale et al., 2021]

Removing ammonia and blocking its reuse may have different consequences. Inhibiting a cancer cell’s ammonia-assimilation pathway could restrict its nitrogen supply while leaving more ammonia outside the cell. That is a mechanistic concern to test, rather than an assumed outcome. Tumor growth, local ammonia, and immune function should be measured together.

The most promising candidate setting would therefore be defined by measurements: substantial local ammonia exposure, an identifiable source or clearance defect, and immune dysfunction that improves when ammonia is reduced. MYC expression alone would be an incomplete way to select a tumor for this approach.

Part V / Making the hypothesis earn its arrows

What Would Demonstrate the Loop?

The decisive experiment must connect both directions in the same system. Comparing MYC levels and ammonia across unrelated tumors cannot establish which controls which.

Perturb MYC in the tumor cells. Measure ammonia production and local exposure early, before substantial cell loss could explain a decline. Track both production and assimilation.
Lower ammonia independently. Determine whether MYC protein and its transcriptional output fall before the tumor shrinks. Use approaches with different mechanisms to help separate ammonia effects from unrelated drug actions.
Test controlled add-back. Restore ammonia while controlling pH, nutrients, and other culture conditions. Ask whether MYC output and immune dysfunction return.
Show sustained coupling. Establish that ammonia-induced MYC activity increases ammonia production sufficiently to maintain the state. Repeat in additional tumor models and include immune-cell killing as a functional readout.

Two outcomes would be informative. If ammonia reduction lowers MYC and restores immune killing, it would support a coupled vulnerability. If MYC remains unchanged while killing improves, ammonia could still be a useful immune target. Its therapeutic relevance does not depend on proving that it came first.

The research opportunity is to identify tumors that depend on the ammonia-rich environment they inhabit—and determine whether changing that environment gives immune cells back the capacity to kill.

Thursday, July 9, 2026

Ammonia-Producing Bacteria of the Human Gut: Location and Cancer Links

Ammonia-Producing Bacteria of the Human Gut

Where They Live Along the GI Tract and How They Relate to Cancer

The gut microbiota generates ammonia by two routes: ureolysis (urease splits host urea into ammonia and CO₂) and proteolytic deamination (fermentation of protein and amino acids). The heaviest producers cluster in the mouth, stomach and distal colon, and several of them are also implicated in cancer, ranging from the WHO Group 1 gastric carcinogen Helicobacter pylori to genotoxin- and toxin-producing colonic species. This post maps the major taxa to their location and to the strength of their cancer evidence, including preclinical in vivo data.

Two Microbial Routes to Ammonia

Understanding why a species produces ammonia matters, because it dictates where in the gut it is active and what it is responding to. Urease-driven producers are limited by urea availability; proteolytic producers are limited by protein and amino-acid supply which is precisely why high-protein and high-fat, dysbiotic diets amplify colonic ammonia. Collectively the gut community hydrolyzes roughly 15–30% of the urea the body synthesizes.

Ammonia Generation Mechanisms

Ureolysis: Microbial ureases hydrolyze host-derived urea diffusing into the lumen into ammonia and carbon dioxide. Classic in H. pylori, Klebsiella, Proteus and oral streptococci.

Amino-Acid Deamination: Proteolytic anaerobes ferment peptides and amino acids e.g. via the Stickland reaction in clostridia, releasing free ammonia. Dominant in the distal colon.

Substrate-Driven: Proteolytic output scales with dietary protein and fat load, so obesity- and diet-induced dysbiosis markedly increases luminal ammonia (millimolar concentrations).

The Map: Taxa, Location and Cancer Evidence

The table below lists the major ammonia-producing taxa in anatomical order, from mouth to colon, with their mechanism and a summary of cancer involvement. Colour tags in the final column indicate the strength of the cancer link.

● Strong / established ● Moderate / preclinical ● Weak / context-dependent

Bacterium	Ammonia mechanism	GI location (mouth → colon)	Cancer involvement (incl. preclinical in vivo)
**Streptococcus salivarius / S. vestibularis**	Urease (~50% of strains; dominant oral ureolytic species)	Mouth — tongue dorsum, oral mucosa, saliva; swallowed onward	● Weak / indirect; oral streptococci associated with oral and esophageal carcinogenesis via inflammation.
*Actinomyces naeslundii*	Urease (urea → ammonia in plaque)	Mouth — dental plaque, biofilm	● Limited; associational only.
*Helicobacter pylori*	Very potent urease (acid-neutralizing survival factor)	Stomach (gastric mucosa)	● Strongest link — WHO Group 1 gastric carcinogen. Ammonia disrupts tight junctions, damages epithelium, drives proliferation; urease itself promotes angiogenesis / HIF-1α. Also MALT lymphoma.
*Streptococcus anginosus*	Proteolytic / ammonia-generating	Mouth → stomach (ectopic gastric colonization)	● Strong preclinical. Promotes gastritis, atrophy, metaplasia and gastric tumorigenesis in mice (Cell, 2024); GSDME-pyroptosis, increased proliferation / invasion.
*Klebsiella pneumoniae*	Potent ureC urease	Small intestine → colon (blooms in dysbiosis / IBD)	● Strong preclinical. ST11 strains exacerbate colitis-associated CRC in mice; T6SS drives inflammation / tumour growth; gut→liver translocation promotes HCC in mice.
*Proteus mirabilis*	Highly potent urease	Colon (dysbiotic expansion)	● Preclinical / associational; enriched in HCC-promoting faecal transplants in mice; pro-inflammatory.
*Morganella morganii*	Urease + proteolytic	Colon (enriched in IBD / CRC)	● Strong preclinical. Produces indolimine genotoxins → DNA double-strand breaks; worsens colon tumorigenesis in gnotobiotic mice, abolished in indolimine-null mutants (Science, 2022).
*Fusobacterium nucleatum*	Proteolytic / amino-acid deamination	Mouth (oral commensal) → ectopic colon	● Strong preclinical + human. Enriched in adenoma / carcinoma; FadA→Wnt/β-catenin, inflammation (IL-1β/6/8, TNF-α), Fap2 immune evasion, chemoresistance.
Bacteroides fragilis — enterotoxigenic (ETBF)	Proteolytic / deamination (+ BFT toxin)	Distal colon	● Strong preclinical. BFT → spermine-oxidase ROS, DNA damage, colon tumorigenesis in mice; increased JMJD2B / stemness, Wnt/β-catenin.
**Bacteroides ovatus & B. vulgatus**	Proteolytic / amino-acid deamination (substrate-driven, not urease)	Distal colon; enriched by high-fat diet / dysbiosis	● Context-dependent, newly mechanistic. In HFD/obesity mouse CRC model, their ammonia disrupts TGF-β tumour suppression — caspase-3 cleaves SPTBN1, ammonia–SPTBN1 adducts trap SMAD3 and drive proinflammatory cytokines (2025). Also BSH → carcinogenic bile acids. But strain-dependent: B. vulgatus can ameliorate HFD obesity; B. ovatus N-methylserotonin can inhibit CRC.
*Clostridium perfringens*	Proteolytic / amino-acid deamination (Stickland reaction); not a urease producer	Small intestine → distal colon (spore-former; blooms in dysbiosis). Essentially absent from mouth / stomach.	● Weak & bidirectional. Mainly a *bacteraemia marker* of pre-existing GI / hepatobiliary cancer** (tumour breaches barrier → translocation), not a proven driver; α-toxin causes damage / inflammation. Conversely, its enterotoxin (CPE) is an anticancer tool — targets claudin-3/4 overexpressed in colon/breast/ovarian tumours; CPE suicide-gene therapy eradicated colon carcinoma in mice.
Other urease⁺ opportunists: Salmonella, Yersinia enterocolitica, Staph. saprophyticus, Ureaplasma urealyticum	Urease	Transient, stomach → colon	● Mostly minimal; chronic Salmonella linked to gallbladder / colon cancer.
Minor proteolytic contributors: Propionibacterium, Lactobacillus, Veillonella	Proteolysis / deamination	Colon (Veillonella also mouth)	● Generally neutral-to-protective; not established drivers.

Reading the Map, Mouth to Colon

Ammonia production is not evenly distributed along the gut. Each compartment has its own dominant producers, substrates and clinical stakes.

Mouth: the first ureolytic station

Saliva is rich in urea, and Streptococcus salivarius (the most abundant, most ureolytic tongue-dorsum coloniser) together with Actinomyces naeslundii hydrolyse it to ammonia, buffering plaque acid. This is largely protective for teeth, and the cancer link is weak but these organisms are swallowed continuously and seed the rest of the tract. Oral Fusobacterium nucleatum also originates here before its consequential migration to the colon.

Key point: an oral commensal in the mouth can become a pathobiont downstream.

Stomach: where ammonia is a proven carcinogenesis promoter

The stomach is the clearest “ammonia → cancer” story. Helicobacter pylori uses an exceptionally potent urease to survive gastric acid, and the resulting ammonium hydroxide is directly implicated in epithelial injury, tight-junction breakdown and hyperproliferation. H. pylori is a WHO Group 1 carcinogen for gastric adenocarcinoma and MALT lymphoma.

Streptococcus anginosus: an oral organism that can ectopically colonise the stomach and independently promotes gastritis, atrophy, metaplasia and gastric tumours in mice, making it an emerging gastric pathobiont.

Small intestine & colon — the ammonia heartland

The distal colon is where undigested protein reaches the densest microbial community, so it is the dominant site of both ureolytic and proteolytic ammonia. Urease-driven Enterobacteriaceae (Klebsiella pneumoniae, Proteus mirabilis, Morganella morganii) bloom under dysbiosis and IBD, while proteolytic anaerobes (Bacteroides, Fusobacterium, Clostridium) ferment amino acids.

This is also where the strongest colorectal cancer mechanisms live: the indolimine genotoxins of M. morganii, the BFT toxin of enterotoxigenic B. fragilis, the FadA/Fap2 machinery of F. nucleatum, and the ammonia–TGF-β axis of B. ovatus/vulgatus.

When Ammonia Itself Is the Weapon: the TGF-β Mechanism

For most of these organisms the demonstrated carcinogenic effector is a toxin or genotoxin, with ammonia as a shared metabolic co-factor. But two settings pinpoint ammonia itself as the driver: the H. pylori gastric story above, and a 2025 study showing that colonic ammonia sabotages the TGF-β tumour-suppressor pathway.

Ammonia → SPTBN1 cleavage → TGF-β failure

In a high-fat-diet mouse model, dysbiosis enriched ammonia-producing Bacteroides ovatus and B. vulgatus. The ammonia they release promotes caspase-3–mediated cleavage of SPTBN1 (βII-spectrin), the adaptor that normally partners with SMAD3.

Normally SPTBN1–SMAD3 travels to the nucleus to switch on tumour-suppressive TGF-β target genes. Ammonia disrupts this: SMAD3 is trapped at the membrane and cytoplasm, and the cleaved SPTBN1 fragments form adducts with ammonia that drive proinflammatory cytokines.

Net effect: tumour-suppressor signalling is switched off while inflammation is switched on — completing a diet → dysbiosis → ammonia → disrupted tumour suppression pathway.

Two Important Caveats

Bacteroides ovatus and B. vulgatus are context-dependent

These species have a genuine Jekyll-and-Hyde literature. Under high-fat/dysbiotic conditions they act as ammonia producers that promote CRC via TGF-β disruption and bile-salt-hydrolase-driven carcinogenic bile acids. Yet in other settings the same species are protective: B. vulgatus supplementation ameliorates high-fat-diet obesity and hyperlipidaemia in mice, and B. ovatus-derived N-methylserotonin can inhibit colorectal cancer.

Their net effect is strain- and context-dependent, determined by substrate load and community composition, not by the species label alone.

Clostridium perfringens: marker, not driver, and a potential therapy

Ammonia route: proteolytic amino-acid deamination via the Stickland reaction. It is not a urease producer, so its output tracks protein load rather than urea.

Cancer link is bidirectional: clinically it is mostly a bacteraemia marker of pre-existing GI or hepatobiliary tumours (a tumour breaches the mucosal barrier and lets the organism translocate), rather than a demonstrated carcinogen.

Why This Matters & What It Doesn't Prove

⚠️ Interpreting the Evidence

The ammonia–cancer relationship is best understood as one thread within a web of protein-fermentation metabolites (alongside phenols, indoles, N-nitroso compounds and hydrogen sulfide) and toxin/genotoxin pathways, not a single-cause story.

Key points:

Only for H. pylori (stomach) and the colonic TGF-β/SPTBN1 axis is ammonia itself firmly identified as a carcinogenesis promoter.
For most colonic species, inflammation, toxins and genotoxins are the proven effectors; ammonia is a co-contributor via cytotoxicity, hyperproliferation and barrier disruption.
Much of the strongest data is preclinical (mouse/gnotobiotic); human causal validation is still developing.
The same species can be harmful or protective depending on strain, diet and community context.

Summary

Two mechanisms, mapped to location: ureolysis dominates the mouth (S. salivarius) and stomach (H. pylori); proteolytic deamination dominates the distal colon.
The clearest ammonia-driven cancer is H. pylori in the stomach (WHO Group 1 carcinogen).
The strongest colonic drivers with preclinical in vivo evidence are M. morganii (indolimine genotoxins), enterotoxigenic B. fragilis (BFT), F. nucleatum (FadA/Fap2) and K. pneumoniae.
A novel ammonia-specific pathway links high-fat diet → B. ovatus/vulgatus → ammonia → caspase-3/SPTBN1 cleavage → TGF-β disruption.
Two organisms defy the pattern: B. ovatus/vulgatus are context-dependent, and C. perfringens is largely a cancer marker whose enterotoxin is being developed as a therapy.

⚠️ Disclaimer

This article is for informational and educational purposes only and does not constitute medical advice. It describes mechanistic and preclinical research and should not be interpreted as diagnosis, treatment recommendations, or a basis for self-testing. Always consult qualified healthcare professionals for personalised medical advice, cancer screening and treatment decisions.

Key considerations:

Presence of a given bacterium does not by itself predict cancer risk; risk is multifactorial.
Much of the cited mechanism comes from animal and cell-based models, not human trials.
Microbiome and dietary interventions should follow evidence-based, medically supervised approaches.
Current cancer screening guidelines remain the standard of care.

References

1. Konieczna I, Zarnowiec P, Kwinkowski M, et al. Bacterial urease and its role in long-lasting human diseases. Curr Protein Pept Sci / PLoS Pathog (Microbial Urease in Health and Disease). PMC4263730

2. Keep calm with ammonia-producing microbiota. Nature Reviews Microbiology. doi:10.1038/s41579-023-00996-x

3. Gut microbiota and dynamics of ammonia metabolism in liver disease. npj Gut and Liver. doi:10.1038/s44355-024-00011-x

4. Shen TD, Albenberg L, Bittinger K, et al. Engineering the gut microbiota to treat hyperammonemia. J Clin Invest. PMC4563680

5. Chen YY, Burne RA. Streptococcus salivarius urease: genetic and biochemical characterization. J Bacteriol. PMC173805

6. Tsujii M, Kawano S, Tsuji S, et al. Ammonia: a possible promotor in Helicobacter pylori-related gastric carcinogenesis. Cancer Lett. doi:10.1016/0304-3835(92)90207-C

7. Helicobacter pylori and Gastric Cancer: Pathogenetic Mechanisms. Int J Mol Sci. PMC9917787

8. Fu K, Cheung AHK, Wong CC, et al. Streptococcus anginosus promotes gastric inflammation, atrophy, and tumorigenesis in mice. Cell. 2024. doi:10.1016/j.cell.2024.01.004

9. Two ST11 Klebsiella pneumoniae strains exacerbate colorectal tumorigenesis in a colitis-associated mouse model. PMC8496539

10. Gut–liver translocation of pathogen Klebsiella pneumoniae promotes hepatocellular carcinoma in mice. Nature Microbiology. doi:10.1038/s41564-024-01890-9

11. Cao Y, Oh J, Xue M, et al. Commensal microbiota from patients with inflammatory bowel disease produce genotoxic metabolites (Morganella morganii indolimines). Science. 2022;378(6618). doi:10.1126/science.abm3233

12. Goodwin AC, Destefano Shields CE, Wu S, et al. Polyamine catabolism contributes to enterotoxigenic Bacteroides fragilis-induced colon tumorigenesis. PNAS. doi:10.1073/pnas.1010203108

13. The Mechanism of Bacteroides fragilis Toxin Contributes to Colon Cancer Formation. PMC7444842

14. Fusobacterium nucleatum in Colorectal Cancer: Ally Mechanism and Targeted Therapy Strategies. PMC11979337

15. Bhowmick K, Yang X, Mohammad T, et al. Microbial metabolite ammonia disrupts TGF-β signaling to promote colon cancer. J Biol Chem. 2025;301(6):108559. PMC12155590

16. Bile salt hydrolase in non-enterotoxigenic Bacteroides potentiates colorectal cancer. Nature Communications. PMC9918522

17. Bacteroides vulgatus ameliorates high-fat diet-induced obesity through modulating intestinal serotonin synthesis and lipid absorption in mice. Gut Microbes. PMC11583587

18. Bacteroides ovatus-derived N-methylserotonin inhibits colorectal cancer via the HTR1D-mediated cAMP-PKA-NF-κB signaling axis. PMC12682799

19. Clostridium perfringens bacteremia associated with colorectal cancer in an elderly woman. PMC7928247

20. Clostridium perfringens Enterotoxin elicits rapid and specific cytolysis of breast carcinoma cells via claudin-3 and -4. PMC1615652

21. Rapid eradication of colon carcinoma by Clostridium perfringens Enterotoxin suicidal gene therapy. BMC Cancer. PMC5307849

22. Ammonia production by human faecal bacteria, and the enumeration, isolation and characterization of bacteria capable of growth on peptides and amino acids. BMC Microbiol. PMID:23312016

Abbreviations: GI: Gastrointestinal; CRC: Colorectal Cancer; HCC: Hepatocellular Carcinoma; IBD: Inflammatory Bowel Disease; HFD: High-Fat Diet; ETBF: Enterotoxigenic Bacteroides fragilis; BFT: Bacteroides fragilis Toxin; CPE: Clostridium perfringens Enterotoxin; BSH: Bile Salt Hydrolase; T6SS: Type VI Secretion System; TGF-β: Transforming Growth Factor Beta; SPTBN1: Spectrin Beta Non-Erythrocytic 1; ROS: Reactive Oxygen Species; MALT: Mucosa-Associated Lymphoid Tissue.

This article integrates review literature, mechanistic studies and preclinical in vivo investigations. Compiled July 2026.

Research Blog - Terms of Use

Pages