Tumour metabolism

Tumour metabolism & the microenvironment

The Tumour’s Supply Problem

How poor blood flow links shortages, waste accumulation and the limits of metabolic therapy.

A tumour can have plenty of blood vessels and still be poorly supplied. Its vessels may be compressed, leaky or unevenly distributed; blood may pass through one region while barely reaching another. The resulting problem runs in two directions: less oxygen and fewer circulating substances reach some cells, while locally produced metabolites are removed less effectively.

This gives a useful connection between two familiar questions: what should be supplied, and what should be reduced? Both depend partly on the circulation. But the connection is not a complete explanation. A tumour’s chemistry also reflects what its cells produce, consume, transport and recycle.

Poor perfusion can link a shortage to an accumulation. It does not make every shortage identical, every accumulation a passive waste product, or every treatment a solution to the same problem.

One circulation, two functions

Delivery in. Clearance out.

What may arrive less effectively

• Oxygen for respiration and oxygen-dependent reactions.
• Nutrients and vitamins, with different uptake and consumption rates.
• Buffering capacity through exchange with circulating blood.
• Treatments, whose penetration depends on their own properties.

What may accumulate locally

• CO₂ and acid equivalents when production outpaces removal.
• Lactate, which may also be taken up and used nearby.
• Ammonia/ammonium, shaped by production, recycling and transport.
• Adenosine, shaped by local generation, uptake and breakdown.

A shared transport constraint; distinct chemical gradients.

Conceptual overview, not a map of measured concentrations. The most hypoxic, acidic and drug-deprived regions need not coincide. [1]

The same blood flow does not mean the same gradient

Oxygen is consumed as it diffuses away from functioning vessels. A drug may instead bind near a vessel, cross its wall slowly, or be removed from the circulation before penetrating deeply. A nutrient may be consumed by one cell and released by another. Distance matters, but it cannot tell the whole story.

Direct measurements illustrate the distinction. In a human-tumour xenograft model, Helmlinger and colleagues measured interstitial oxygen and pH at high spatial resolution and found that their local gradients did not correlate. That does not make perfusion irrelevant. It shows why oxygen cannot be used as a substitute measurement for acidity. [1]

Acidity also needs more than a blocked-drain explanation. Carbonic anhydrase IX accelerates the reversible conversion between CO₂, water, bicarbonate and protons; it does not create carbon dioxide. Experiments in mouse xenografts show that its expression can change extracellular pH. The final pH reflects this chemistry alongside cellular acid handling, buffers and exchange with surrounding tissue. It cannot reveal, by itself, how much acidification came from each source. [2]

What can be supplied—and what delivery actually proves

Oxygen matters both as a reactant and as a regulator of hypoxia responses. Yet increasing oxygen in blood and increasing oxygen inside a poorly perfused tumour are different achievements. A convincing intervention needs evidence that the relevant tissue became better oxygenated. Clinical imaging during cediranib treatment provides one example of how perfusion and oxygenation can be assessed together. [3]

Bicarbonate is part of the extracellular buffering system. In mouse studies, oral bicarbonate increased tumour extracellular pH and reduced metastasis in some models, with different results in others. These findings establish a preclinical effect under particular conditions. They do not establish that raising systemic bicarbonate reliably corrects human tumour acidity or improves survival. [4]

Vitamins and cofactors should not be treated as one interchangeable supply category. A shared route through the blood does not establish a shared tissue concentration, deficiency or response to supplementation. Ascorbate is of interest because there are both human tissue measurements and a small randomised treatment trial.

Ascorbate: a delivery problem that has been measured

In a small presurgical study of colon cancer, nine patients received intravenous ascorbate and six served as controls. Tumour ascorbate increased after infusion, while concentrations were lower toward the tumour centre in both groups. This is direct human evidence of spatial variation and improved tissue delivery. It is not a complete map of ascorbate against local blood flow, and it did not establish a survival benefit. [5]

At pharmacological concentrations, ascorbate can generate hydrogen peroxide. Experimental susceptibility depends on redox-active iron and the cell’s ability to handle peroxide, among other factors. Oxygen availability is relevant to that chemistry, but hypoxia does not establish that labile iron is absent. Nor can all cancer cells be assumed to have weaker antioxidant defences than normal cells. [6]

A promising trial, with a precise boundary

A randomised phase 2 trial in metastatic pancreatic cancer assigned 36 patients; 34 received study treatment. Adding intravenous pharmacological ascorbate to gemcitabine and nab-paclitaxel produced median overall survival of 16.0 versus 8.3 months and progression-free survival of 6.2 versus 3.9 months. The publication reports hazard ratios: 0.46 for overall survival (90% CI 0.23–0.92) and 0.43 for progression-free survival (90% CI 0.20–0.92). [7]

The trial stopped early after an interim analysis. Its small size, early stopping and phase 2 statistical design make the magnitude of benefit uncertain. It supports further testing of this particular intravenous combination, not extrapolation to oral vitamin C or other cancers.

A consequence can still be a worthwhile target

It is tempting to divide metabolic targets into two lists: consequences of poor perfusion and intrinsic needs of cancer cells. That distinction can organise a discussion, but it cannot rank treatments on its own. A downstream mediator can help sustain disease; an intrinsic dependency can vary with the environment or be bypassed.

Adenosine illustrates why the mechanisms need separating. CD39 converts extracellular ATP/ADP toward AMP, and CD73 converts AMP to adenosine. Hypoxia can strengthen this pathway, but “HIF turns on all of it” is too simple: direct experimental work identified HIF-dependent CD73 regulation and Sp1-dependent CD39 induction in different systems. HIF-dependent regulation of the A2B receptor has also been demonstrated. These mechanisms are context-dependent, not a universal tumour switch. [8] [9] [10]

In the randomised phase 2 ARC-9 cohort B, 112 patients with previously treated metastatic colorectal cancer received either etrumadenant, zimberelimab, mFOLFOX-6 and bevacizumab or regorafenib. Median overall survival was 19.7 versus 9.5 months (HR 0.37, 95% CI 0.22–0.63); progression-free survival was 6.2 versus 2.1 months (HR 0.27, 95% CI 0.17–0.43). [11]

That is evidence for the four-part regimen against its comparator. The study cannot isolate the contribution of etrumadenant, the adenosine-receptor antagonist. Its result therefore supports investigating adenosine-directed combinations without proving that adenosine blockade alone produced the survival gain.

Ammonia is also more than an uncleared by-product. Isotope-tracing experiments showed that breast cancer cells can recycle ammonia nitrogen into amino acids and tumour biomass. Its local concentration therefore reflects both production and use, as well as removal. Calling it “waste” describes its origin in one reaction, not everything it does in the tissue. [12]

Glutamine and cysteine remain important experimental vulnerabilities, but not universally fixed ones. In mouse pancreatic cancer, genetic disruption of cystine uptake or pharmacological cysteine/cystine depletion induced ferroptosis and inhibited tumour growth. Separately, the glutamine-antagonist prodrug JHU083 suppressed tumour metabolism and supported antitumour T-cell responses in mice. These are distinct interventions with preclinical evidence; neither establishes a general nutrient-restriction treatment for patients. [13] [14]

Three questions for every target

Cause, persistence and vulnerability

What produces it?

Cellular metabolism, gene regulation and local substrate availability.

What keeps it there?

Production versus consumption, transport, diffusion and vascular removal.

What depends on it?

Cancer-cell survival, immune activity or a process that can be disrupted selectively.

An interpretive framework, not a validated treatment ranking. The same molecule can belong in all three questions.

Where the interventions interact

The useful question is not whether two agents sound complementary. It is whether one changes the conditions the other needs—and whether that change helps the tumour, the immune system or the patient.

Potential interactions: mechanisms to test, not a combination protocol

Pair	What could change	What can be concluded
Buffering & ammonia	At fixed total ammonia, a higher pH increases the uncharged NH3 fraction.	An equilibrium prediction; it does not establish net tissue exposure or immune harm. [15]
Buffering & MCT transport	A change in pH alters proton-coupled transport, alongside substrate gradients.	Effects depend on the actual substrate and both sides of the membrane. An MCT inhibitor is not necessarily an MCT-transported drug. [16]
Ascorbate & hypoxia	Oxygen availability can constrain peroxide-generating chemistry.	Measure delivery, oxygen and redox response together; hypoxia alone does not predict failure. [6]
Glucose & glutamine targeting	Fuel use may adapt, but inhibiting glutamine metabolism can also suppress glycolysis.	Neither universal compensation nor a preferred sequential schedule follows from pathway diagrams. [14]
Ascorbate & cystine targeting	Oxidative stress and impaired antioxidant defence may interact.	Related combinations already have laboratory evidence; formulation and cell context matter. [17] [18]
Oxygenation & hypoxia responses	Improved oxygen delivery may change hypoxia-dependent signalling.	Confirm tissue oxygenation and the relevant response; an upstream change does not guarantee all downstream effects disappear. [3]

The ammonia arithmetic—and its limit

For the NH₄⁺/NH₃ equilibrium, the uncharged fraction is 1 / (1 + 10^{pKₐ − pH}). Using an illustrative fixed total concentration of 5 mM and pKₐ = 9.25 gives approximately 14 µM NH₃ at pH 6.7 and 70 µM at pH 7.4. That is about a fivefold increase.

Those numbers reproduce a conditional calculation, not an observed response to bicarbonate treatment. The pKₐ varies with conditions; using approximately 9.15 for physiological temperature and osmolality gives about 18 and 87 µM instead. In tissue, total ammonia need not remain fixed while pH changes. Transport, metabolism and removal can all change too. NH₄⁺ also has protein-mediated transport routes; it is not absolutely membrane-impermeant. [15]

To infer impaired antigen processing would require additional measurements of cellular uptake, organelle pH and immune function. The calculation identifies a variable worth measuring. It cannot establish that reducing tumour acidity harms immunity.

Ascorbate and cystine: a question already being tested

The proposed pairing is not an untouched idea. In a 2022 study, inhibiting cystine uptake with imidazole ketone erastin only mildly enhanced pharmacological ascorbate activity in the tested cells. That work also distinguished ascorbate-induced death from classic GPX4-dependent ferroptosis. [17]

A separate 2023 study reported sensitisation to system x_c⁻ inhibition using dehydroascorbic acid, the oxidised form of vitamin C. This is relevant preclinical evidence, but dehydroascorbic acid and infused ascorbate are not interchangeable experimental exposures. The next question is which specific intervention works, in which tumour, with what selectivity—not simply whether two agents affect glutathione. [18]

Can improving blood flow address both sides?

It can potentially improve delivery and removal together. This is the appeal of vascular normalisation: improving the function of abnormal tumour vessels, rather than assuming that fewer vessels always means a better outcome.

There is human evidence consistent with this idea. In a study of glioblastoma patients receiving cediranib with chemoradiation, the subgroup whose tumour perfusion increased also showed improved oxygenation and longer survival. This association supports the mechanism, but it does not prove that increased perfusion alone caused the survival difference. [3]

The treatment result still has to be tested separately. A phase 3 trial in recurrent glioblastoma did not show a significant progression-free survival improvement with cediranib, alone or with lomustine, against lomustine. These studies involved different settings and cannot be substituted for one another; together they illustrate the distinction between an informative physiological response and a reliably effective regimen. [19]

A 2026 review by Subudhi, Lee and Jain describes the opportunities and limitations of vascular normalisation after 25 years of research. Dose and timing matter: excessive vessel pruning can worsen hypoxia, and tumours can use alternative vascular-growth pathways or co-opt existing vessels. There is no universal intervention that simply restores a normal circulation in every tumour. [20]

Improving perfusion is also not the only way to change several environmental stresses at once. In the JHU083 mouse experiments, suppressing tumour metabolism reduced hypoxia and acidosis. Changing demand and production can therefore alter the environment even without demonstrating that the vascular defect itself was repaired. [14]

What this changes about the question

The strongest conclusion is a way to evaluate interventions, rather than a list of preferred agents. A treatment can fail because too little reaches its target, because the local chemistry limits its action, because the proposed vulnerability is absent, or because the tumour adapts. These explanations call for different experiments.

• Measure arrival: did the agent reach the relevant tumour region at an active concentration?
• Measure action: did oxygen, pH, a metabolic flux or the intended target actually change?
• Measure selectivity: were cancer cells affected differently from immune and normal cells?
• Measure benefit: did those changes improve tumour control or patient outcomes?

There is no general rule that targeting a consequence must produce only a temporary benefit, or that targeting a dependency has no ceiling. Both propositions need evidence in the disease and treatment being considered.

The enduring point is that delivery and biology must be studied together. Poor blood flow can help create a hostile environment and limit the treatments intended to change it. Understanding that link is useful precisely when it leads to measurements—not when it replaces them.

Evidence note: targeted literature review checked on 16 October 2026, not a systematic review. Human treatment trials, tissue measurements, animal experiments and mechanistic predictions are identified separately. This article discusses research; it does not establish a treatment or dietary-restriction protocol.

References

1. Helmlinger G, et al. Interstitial pH and pO₂ gradients in solid tumors in vivo: high-resolution measurements reveal a lack of correlation. Nature Medicine (1997). Spatial measurements in a human-tumour xenograft model.
2. Carbonic anhydrase IX is a pH-stat that sets an acidic tumour extracellular pH in vivo. British Journal of Cancer (2018). Experimental pH measurements in mouse xenografts.
3. Batchelor TT, et al. Improved tumor oxygenation and survival in glioblastoma patients who show increased blood perfusion after cediranib and chemoradiation. PNAS (2013). Human imaging and survival associations during treatment.
4. Robey IF, et al. Bicarbonate increases tumor pH and inhibits spontaneous metastases. Cancer Research (2009). Mouse experiments; effects varied by model.
5. Dachs GU, et al. Vitamin C administration by intravenous infusion increases tumor ascorbate content in patients with colon cancer: a clinical intervention study. Frontiers in Oncology (2021). Small randomised presurgical tissue study.
6. Schoenfeld JD, et al. O₂•− and H₂O₂-mediated disruption of Fe metabolism causes the differential susceptibility of NSCLC and GBM cancer cells to pharmacological ascorbate. Cancer Cell (2017). Mechanistic experiments with early clinical investigation.
7. A randomized trial of pharmacological ascorbate, gemcitabine, and nab-paclitaxel for metastatic pancreatic cancer. Redox Biology 77, 103375 (2024). Small phase 2 trial; early stopping; reported 90% confidence intervals.
8. Synnestvedt K, et al. Ecto-5′-nucleotidase (CD73) regulation by hypoxia-inducible factor-1 mediates permeability changes in intestinal epithelia. Journal of Clinical Investigation (2002). Experimental hypoxia signalling; not a cancer efficacy trial.
9. Central role of Sp1-regulated CD39 in hypoxia/ischemia protection. Blood (2009). Experimental CD39 regulation.
10. Kong T, et al. HIF-dependent induction of adenosine A2B receptor in hypoxia. FASEB Journal (2006). Experimental receptor regulation.
11. The randomized phase II ARC-9 study of etrumadenant-based therapy versus regorafenib in patients with previously treated metastatic colorectal cancer. Clinical Cancer Research (2026). Randomised phase 2 comparison of regimens; no isolated etrumadenant effect.
12. Spinelli JB, et al. Metabolic recycling of ammonia via glutamate dehydrogenase supports breast cancer biomass. Science (2017). Cell and tumour-model isotope tracing.
13. Badgley MA, et al. Cysteine depletion induces pancreatic tumor ferroptosis in mice. Science (2020). Genetic and pharmacological mouse experiments.
14. Leone RD, et al. Glutamine blockade induces divergent metabolic programs to overcome tumor immune evasion. Science (2019). Preclinical glutamine-antagonist experiments.
15. Mechanisms of ammonia and ammonium transport by rhesus-associated glycoproteins. American Journal of Physiology–Cell Physiology (2015). See also Ammonia transporters and their role in acid-base balance (2017), for the physiological equilibrium convention. Transport experiments and a physiology review; not evidence of bicarbonate-induced tumour immune dysfunction.
16. Human monocarboxylate transporters accept and relay protons via the bound substrate for selectivity and activity at physiological pH. PNAS Nexus (2023). Transport mechanism experiments.
17. Jankowski CSR, Rabinowitz JD. Selenium modulates cancer cell response to pharmacologic ascorbate. Cancer Research (2022). Preclinical redox and combination experiments.
18. Dehydroascorbic acid sensitizes cancer cells to system x_c⁻ inhibition-induced ferroptosis by promoting lipid droplet peroxidation. Cell Death & Disease 14, 637 (2023). Preclinical combination study using oxidised vitamin C.
19. Batchelor TT, et al. Phase III randomized trial comparing the efficacy of cediranib as monotherapy, and in combination with lomustine, versus lomustine alone in patients with recurrent glioblastoma. Journal of Clinical Oncology (2013). Randomised phase 3 trial.
20. Subudhi S, Lee S, Jain RK. Targeting angiogenesis: lessons from 25 years of normalizing tumor vasculature. Cell 189, 2379–2415 (2026). Review and synthesis of vascular-normalisation research.