Beyond the Warburg Effect - Cancer as a Metabolic Ecosystem
Cancer metabolism is far more complex—and more fascinating—than the familiar idea that tumors simply “feed on sugar.” New research reveals tumors as dynamic metabolic ecosystems: cancer cells can switch fuels, exchange nutrients with neighboring cells, manipulate immune cells, reshape gene activity, and even transfer mitochondria between cells to survive and adapt. This emerging picture helps explain why some metabolic treatments work spectacularly in certain cancers yet fail in others, and why the future may lie not in starving cancer with a single diet or drug, but in identifying the specific metabolic dependencies and escape routes each tumor cannot live without.
The Overview
The Warburg effect remains one of the central observations in cancer biology, but it no longer describes the whole metabolic life of a tumor.
Many cancer cells use glycolysis heavily while continuing to depend on mitochondria for energy, biosynthesis, antioxidant protection, signaling, and survival.
A tumor does not have a single metabolism. Different cells and different regions of the same tumor can use different fuels and can change those fuels over time.
Glucose and glutamine remain important, but cancer cells may also use lactate, fatty acids, acetate, ketones, amino acids, and materials supplied by surrounding cells.
Lactate is not merely metabolic waste. It can be transported between cells, burned as fuel, alter acidity, influence immunity, encourage blood-vessel growth, and modify gene activity.
Fibroblasts, immune cells, blood vessels, fat cells, and other neighboring cells can become active participants in the tumor economy, supplying nutrients or creating conditions that protect cancer cells.
The tumor microenvironment is a metabolic battleground. Cancer cells can deprive immune cells of nutrients and surround them with metabolites that reduce their ability to attack.
Recent research shows that whole mitochondria can move between cancer and immune cells, changing the energy capacity and behavior of both.
Genes and metabolism should not be treated as competing explanations. Mutations reprogram metabolism, while metabolites influence gene expression, DNA repair, cell identity, and the selection of new mutations.
Metabolic treatment is most likely to succeed when it targets a defined dependency in a particular tumor and limits the cancer’s ability to switch fuels or obtain help from its surroundings.
How Tumors Exchange Fuels, Manipulate Immunity, Reprogram Cell Behavior, and Adapt to Treatment
For nearly a century, the Warburg effect has offered one of the most compelling clues to the nature of cancer. Otto Warburg observed that cancer cells consume large amounts of glucose and convert much of it into lactate even when oxygen is available. Normal textbooks had taught that cells use fermentation mainly when oxygen is scarce. Cancer cells appeared to be choosing a less efficient route even when the more efficient route remained open. That observation was so consistent that it eventually became the biological basis for fluorodeoxyglucose positron emission tomography, or FDG-PET, which detects areas of unusually high glucose uptake.
Warburg believed that impaired cellular respiration, centered in the mitochondria, was the fundamental cause of cancer. Later generations of researchers shifted their attention toward mutations in oncogenes and tumor-suppressor genes. In recent decades, metabolism returned to the center of cancer research, and the debate was often framed as a contest: is cancer primarily genetic, or is it primarily metabolic?
The newer evidence suggests that this is no longer the most useful question. Cancer is genetic, metabolic, epigenetic, immunologic, and ecological at the same time. A mutation can change how a cell uses glucose. A metabolic change can alter which genes are active. A shortage of oxygen can select for cells that survive through glycolysis. Lactate produced by one cell can become fuel for another. Fibroblasts can release nutrients that protect cancer cells during starvation. Metabolic competition can weaken T cells. Treatment can eliminate one population while leaving behind cells that use a different energy strategy.
The most important advance, therefore, is not the discovery of another fuel used by cancer. It is the realization that a tumor behaves as a metabolic ecosystem. A tumor is not simply a mass of identical malignant cells. It is a changing community composed of cancer cells, immune cells, connective-tissue cells, blood vessels, extracellular matrix, signaling molecules, and regions with very different supplies of oxygen and nutrients. These components compete, cooperate, exchange resources, and change one another.
This perspective preserves Warburg’s great insight: altered metabolism is fundamental to cancer. At the same time, it moves beyond the idea that every cancer cell has a uniformly damaged energy system or that cancer can be treated simply by depriving it of one fuel. The deeper story is about flexibility, specialization, communication, and adaptation. Understanding that story helps explain why metabolic therapies can be highly effective in a specific context yet disappointing when applied broadly. It also points toward a more precise future in which the metabolic organization of an individual tumor can help guide treatment.
Warburg’s Great Insight - and What It Did Not Mean
The Warburg effect is often summarized by saying that cancer cells stop using mitochondria and switch to glycolysis. That description is easy to remember, but it is usually too simple. Glycolysis is the pathway that breaks glucose into pyruvate in the fluid portion of the cell. Pyruvate can then enter mitochondria and be oxidized, or it can be converted into lactate. Many cancer cells send an unusually large proportion of glucose toward lactate even when oxygen is present. This is why the process is called aerobic glycolysis.
At first glance, this appears wasteful. Complete oxidation of glucose through the mitochondria can produce far more ATP than glycolysis alone. ATP is the cell’s immediately usable energy currency. If energy efficiency were the only goal, the cancer cell’s strategy would make little sense.
A rapidly dividing cell, however, does not need only ATP. It must duplicate membranes, proteins, DNA, RNA, organelles, and the chemical systems that protect it from stress. Glycolysis allows glucose-derived carbon to be diverted into pathways that make nucleotides, amino acids, lipids, and antioxidant molecules. It can produce ATP quickly, regenerate NAD+, and continue under fluctuating oxygen conditions. In this context, the cancer cell is not merely choosing an inferior engine. It is reorganizing metabolism around growth, repair, and survival.
This does not mean mitochondrial dysfunction is unimportant. Many cancers contain mitochondrial mutations, abnormal mitochondrial structure, disrupted electron transport, altered fusion and fission, or impaired quality control. Mitochondrial damage can increase reactive oxygen species, destabilize the genome, alter cell signaling, and encourage compensatory glycolysis. In some tumors, respiratory impairment may be an early and powerful driver.
Yet many tumors retain considerable mitochondrial activity. Some use oxygen efficiently. Some depend heavily on oxidative phosphorylation, the mitochondrial process that couples nutrient oxidation to ATP production. Others use mitochondria less for ATP than for producing aspartate, citrate, iron-sulfur clusters, heme, and other essential materials. Even strongly glycolytic tumors may require functioning mitochondria to proliferate.
The modern view is therefore less absolute. The Warburg effect does not prove that mitochondria are universally inactive, nor does preserved oxygen consumption prove that metabolism is normal. Cancer cells may operate both glycolysis and oxidative metabolism at high rates. They may use one pathway for ATP, another for biosynthesis, and another for managing oxidative stress. The important abnormality is not simply that one pathway is turned off and another is turned on. It is that the entire network is reorganized around the priorities of malignancy.
Why a Fast but Inefficient Pathway Can Be Useful
A useful way to understand aerobic glycolysis is to distinguish efficiency from throughput. A small workshop may use every piece of material with great efficiency but produce only a few items each day. A busy construction site may accept waste because materials must move rapidly through many parallel processes. Cancer resembles the construction site. It needs energy, but it also needs a continuous stream of carbon skeletons, reducing power, and molecular parts.
The pentose phosphate pathway, for example, branches from glycolysis and produces ribose for nucleotides as well as NADPH, which supports fatty-acid synthesis and antioxidant defenses. Glycolytic intermediates can contribute to serine and glycine production. Citrate exported from mitochondria can support lipid synthesis. Glutamine can refill the tricarboxylic acid cycle and provide both carbon and nitrogen. What appears to be a single pathway is actually an interconnected network that can be redirected according to the cell’s needs.
Aerobic glycolysis also helps cells survive uncertainty. Oxygen delivery within a tumor can change as blood vessels collapse, reopen, or grow irregularly. A cell already equipped for high glycolytic activity is less vulnerable to sudden hypoxia. In that sense, the Warburg effect may serve as metabolic insurance as well as a growth program.
A Tumor Does Not Have One Metabolism
One of the most important changes in cancer research has come from the ability to examine metabolism at much finer resolution. Older studies often ground up a tumor, measured the average levels of enzymes or metabolites, and described the result as the metabolism of that cancer. The average was useful, but it concealed differences among cells.
A solid tumor contains regions close to blood vessels and regions far from them. Some areas receive oxygen and glucose relatively easily. Others are intermittently starved. The outer edge of a tumor may interact with healthy tissue, nerves, fat cells, and immune cells in ways that do not occur at the center. A cell preparing to invade may have different requirements from a cell dividing rapidly in the primary mass. A cell that survived radiation may not resemble the cell that was present before treatment.
New spatial metabolomics and isotope-tracing methods can now follow labeled nutrients through individual cells while preserving their location within tissue. In 2024, researchers used spatial single-cell isotope tracing to show substantial differences in glucose-derived fatty-acid synthesis among neighboring cancer cells. The significance was not limited to lipid production. It demonstrated that metabolic heterogeneity can exist at the level of individual cells and can be organized spatially within the tumor.
This helps resolve many apparent contradictions in cancer metabolism. One study may find that a tumor is strongly glycolytic, while another finds substantial oxidative phosphorylation. Both may be correct. One cell population may export lactate while another imports and oxidizes it. One region may depend on glutamine while another obtains carbon from fatty acids. The tumor may change its dominant strategy after treatment.
Metabolic heterogeneity also affects diagnosis. A biopsy samples only a small area. A metabolic enzyme that appears low in that sample may be abundant elsewhere. A PET scan can show regional glucose uptake but may miss cells relying on other fuels. Blood metabolites represent a mixture of tumor metabolism, normal tissue metabolism, diet, liver function, muscle activity, and the response of the whole body. Precision metabolic oncology will require ways to combine these sources of information rather than assuming that any one measurement captures the entire tumor.
Metabolic States Can Change Without New Mutations
Cancer cells do not always need a new mutation to change their metabolism. Enzyme activity can be altered by phosphorylation, acetylation, oxidation, nutrient availability, oxygen sensing, or changes in gene expression. Mitochondria can be divided, fused, removed through mitophagy, or produced in greater numbers. Transporters can move to the cell membrane. A cell can enter a slow-growing state and later return to active proliferation.
This flexibility is especially important during treatment. Chemotherapy, targeted drugs, radiation, and immune attack all change the environment. The surviving cells may use more oxidative phosphorylation, increase fatty-acid oxidation, strengthen antioxidant systems, or temporarily slow growth. These changes can be reversible. The tumor has not necessarily acquired permanent genetic resistance, but it has entered a state that allows it to endure.
This is one reason metabolism is both attractive and difficult as a target. Metabolic programs are necessary for survival, but they are also adaptable. The key therapeutic question becomes not merely, “Which pathway is active?” but “Which pathway is essential under these particular conditions, and what alternatives remain available if it is blocked?”
Mitochondria Remain Central to Cancer
Mitochondria are often called the power plants of the cell, but this description understates their importance. They are also manufacturing centers, redox regulators, calcium buffers, immune-signaling platforms, and decision points for programmed cell death. A cancer cell may reduce one mitochondrial function while depending intensely on several others.
The tricarboxylic acid cycle, also called the TCA or Krebs cycle, receives carbon from glucose, glutamine, fatty acids, lactate, acetate, and several amino acids. Its intermediates are continually withdrawn to make lipids, nucleotides, heme, amino acids, and other molecules. To keep the cycle operating, cancer cells must replace those intermediates through a process called anaplerosis. Glutamine often serves this role, but it is not the only source.
Mitochondria also help control reactive oxygen species. Cancer cells commonly operate under higher oxidative stress than normal cells because of rapid metabolism, oncogenic signaling, inflammation, hypoxia, and damaged cellular components. Moderate levels of reactive oxygen species can support proliferation and signaling. Excessive levels can damage membranes, proteins, and DNA or trigger cell death. Cancer cells must therefore maintain a narrow balance: enough oxidation to support signaling and growth, but not enough to become lethal.
This balance creates dependencies on glutathione, thioredoxin, NADPH, cysteine uptake, iron handling, and mitochondrial quality control. A metabolic therapy may work not by starving the cell of ATP, but by overwhelming its ability to contain oxidative damage.
Some of the cells most difficult to eliminate may be especially dependent on mitochondria. Cancer stem-like cells, metastatic cells, and drug-tolerant persister cells often show increased oxidative phosphorylation in at least some tumor types. Persister cells are a small population that survives treatment without necessarily carrying stable resistance mutations. They may slow down, improve mitochondrial quality control, and increase antioxidant capacity. Research has implicated PINK1-mediated mitophagy - the selective removal of damaged mitochondria - in maintaining oxidative phosphorylation and redox stability in these survivors.
The lesson is not that oxidative phosphorylation replaces glycolysis as the true cancer metabolism. The lesson is that tumors can use either, or both, and can redistribute their dependence according to circumstance. A treatment directed only at glycolysis may leave oxidative cells behind. A mitochondrial inhibitor may strongly affect one population while selecting for cells that increase glycolysis. Combination strategies must therefore be designed around actual dependencies rather than broad labels.
Mitochondria as a Source of Signals
Mitochondria communicate with the nucleus and the rest of the cell. Changes in ATP, NADH, reactive oxygen species, calcium, acetyl-CoA, alpha-ketoglutarate, succinate, and other metabolites can change enzyme activity and gene expression. Mitochondrial stress can activate inflammatory pathways. Mitochondrial DNA released into the cytoplasm can be interpreted as a danger signal and activate the cGAS-STING system.
These signals can have different effects depending on context. STING activation may help immune cells recognize cancer, but chronic or misdirected activation can also encourage inflammation, immune escape, or metastasis. This duality is common in cancer biology. A pathway that protects the body under one condition can be recruited by a tumor under another.
Metabolic Plasticity: Cancer Changes Fuels
The phrase “sugar feeds cancer” contains an important truth but can easily become misleading. Glucose is essential to many tumors, and excessive insulin signaling, obesity-associated inflammation, and high circulating nutrients can create conditions favorable to cancer. Yet glucose is not the only usable fuel, and the body cannot simply remove glucose from the circulation without threatening the brain, red blood cells, and other normal tissues.
Cancer cells may obtain carbon and energy from glutamine, lactate, fatty acids, acetate, ketone bodies, branched-chain amino acids, serine, glycine, methionine, asparagine, and materials scavenged from proteins or dead cells. The importance of each source varies by tissue, mutation, stage, and environment.
Glutamine illustrates the complexity. It provides nitrogen for nucleotides and amino acids, carbon for the TCA cycle, and material for glutathione production. Some tumors become strongly dependent on glutaminase, the enzyme that converts glutamine to glutamate. Yet other tumors synthesize glutamine, obtain it from neighboring cells, or bypass glutaminase through alternative pathways. A glutamine inhibitor can therefore be powerful in a selected metabolic setting and ineffective in another.
Fatty acids are equally diverse in their roles. Cancer cells can synthesize them, import them from the circulation, receive them from nearby adipocytes, store them in lipid droplets, use them to make membranes, or oxidize them in mitochondria. Lipid droplets can protect cells by temporarily storing potentially toxic fats. In other settings, polyunsaturated fatty acids make membranes vulnerable to lipid peroxidation and ferroptosis, an iron-dependent form of cell death.
Acetate can become important in hypoxic or nutrient-poor environments. Some tumors use extracellular proteins through macropinocytosis, a process resembling cellular drinking, and break those proteins into amino acids. Pancreatic cancers growing in a dense, poorly perfused environment may obtain alanine released by pancreatic stellate cells, reducing their dependence on glucose-derived carbon.
This flexibility explains why blocking a single nutrient often slows growth without producing lasting control. The tumor may lower its energy use, increase scavenging, receive metabolites from surrounding cells, or switch to another substrate. Effective metabolic treatment will often require either simultaneous pressure at several points or a strategy that blocks the adaptation itself.
The Difference Between a Preferred Fuel and an Essential Fuel
A tumor may consume a nutrient rapidly without being absolutely dependent on it. High uptake can reflect abundance rather than necessity. Conversely, a nutrient used in smaller quantities may be essential because it supplies a molecule that cannot be replaced.
This distinction is crucial in drug development. Laboratory cells are commonly grown in media containing far more glucose, glutamine, oxygen, and growth factors than they encounter in a human tumor. A pathway that appears essential in culture may be bypassed in living tissue. The reverse can also occur: interactions with immune or stromal cells may create a dependency that is invisible in isolated cancer cells.
Human isotope-tracing studies are helping address this problem by infusing labeled nutrients into patients before surgery and examining how those nutrients are used in the actual tumor. These studies have shown, among other findings, that human tumors may oxidize lactate and other circulating fuels extensively. The future of metabolic therapy depends on understanding metabolism in the patient, not only in a dish.
Lactate: From Waste Product to Metabolic Currency
Lactate was once treated mainly as the end product of glycolysis and a marker of insufficient oxygen. It is now understood as one of the body’s major circulating carbon carriers. Muscles, red blood cells, skin, brain, and other tissues continually produce and consume lactate. The liver can convert it back into glucose, and mitochondria can oxidize it after it is converted to pyruvate.
Cancer cells often produce large amounts of lactate, but that lactate does not simply accumulate as useless waste. It can be exported through monocarboxylate transporters such as MCT4 and imported through transporters such as MCT1. In a metabolically organized tumor, hypoxic glycolytic cells may release lactate while better-oxygenated cells near blood vessels import it and use it in the TCA cycle. Human lung-tumor studies have shown that lactate can contribute substantially to tumor metabolis.
This exchange creates a form of metabolic cooperation. The glycolytic cell regenerates NAD+ and removes acid-associated lactate. The oxidative cell receives a carbon source and may spare glucose for other uses. Blocking lactate transport can therefore disrupt not one cell but a relationship between populations.
Lactate also changes the environment. When exported with protons, it contributes to extracellular acidity. An acidic environment can activate enzymes that degrade surrounding tissue, alter cell adhesion, reduce the effectiveness of some immune cells, and favor invasion. Acid is not produced by lactate alone, but high glycolytic flux and proton export are important contributors.
The immune effects are especially important. High lactate concentrations and low pH can reduce T-cell proliferation, cytokine production, and killing capacity. They can encourage macrophages to adopt tissue-repair and immunosuppressive behaviors that the tumor exploits. In preclinical models, blocking MCT4 reduced lactate-driven immune suppression and improved the activity of immune checkpoint therapy.
Lactate Can Influence Gene Activity
The discovery of histone lactylation added another dimension. Histones are proteins around which DNA is wrapped. Chemical modifications to histones help determine which regions of DNA are accessible for gene expression. In 2019, researchers demonstrated that lactate-derived chemical groups could be added to histones, linking glycolytic metabolism directly to the control of gene activity.
Lactylation has since been observed in cancer cells, macrophages, endothelial cells, and other components of the tumor microenvironment. Its effects are context-dependent, but many studies suggest that it can support repair-like macrophage behavior, angiogenesis, proliferation, or resistance. A 2025 melanoma study found that lactate-driven histone lactylation enhanced the angiogenic behavior of tumor-associated endothelial cells through the IL-33/ST2 pathway.
The larger implication is more important than any single pathway. A metabolite produced during energy processing can become an instruction that changes cellular identity. Metabolism is therefore not merely the machinery beneath cell behavior. It is part of the information system that directs that behavior.
The Tumor’s Neighbors Become Part of Its Metabolism
A tumor contains many cells that are not malignant. Cancer-associated fibroblasts produce connective tissue and remodel the extracellular matrix. Endothelial cells form blood vessels. Macrophages, lymphocytes, neutrophils, and other immune cells enter the tissue. Adipocytes may surround tumors in the breast, ovary, pancreas, and other organs. Nerves and glial cells can influence growth. These cells have their own metabolic needs, but cancer can recruit them into a shared economy.
Fibroblasts are especially important. Cancer cells can stimulate fibroblasts to increase autophagy, glycolysis, amino-acid production, or extracellular-matrix breakdown. The fibroblasts may then release lactate, alanine, glutamine, lipids, or other materials that cancer cells use. In pancreatic cancer, stellate-cell-derived alanine can feed tumor metabolism when glucose and serum nutrients are limite. Collagen itself can become a nutrient reserve: cancer cells and stromal cells may degrade the matrix and use the released proline and other amino acids.
This is not simply passive feeding. The tumor can create the conditions that cause neighboring cells to provide resources. Signals from cancer cells alter fibroblast gene expression. Treatment may intensify the exchange. Hormone deprivation, radiation, or chemotherapy can change stromal metabolism in ways that unintentionally protect residual cancer cells.
Adipocytes can supply fatty acids and inflammatory signals. This is especially relevant when tumors grow within or spread to fat-rich tissues. Nearby fat cells may release lipids that cancer cells oxidize for energy or use to build membranes. Obesity can add systemic changes - insulin resistance, altered adipokines, chronic inflammation, and increased nutrient availability - that interact with local tumor metabolism.
Macrophages also participate in nutrient exchange and tissue remodeling. Some engulf dead cells and release metabolites. Others produce growth factors, vascular signals, or enzymes that reshape the matrix. Their behavior is influenced by oxygen, lactate, lipids, iron, and amino acids. Rather than existing outside the metabolic theory of cancer, immune and stromal cells are part of it.
Why the Ecosystem Perspective Changes Treatment
A drug may successfully inhibit a metabolic pathway in isolated cancer cells but fail in a tumor because surrounding cells provide the missing product. A dietary intervention may lower one circulating fuel while the tumor increases scavenging or receives substrates from the stroma. A therapy that kills many cells may release nutrients that help the survivors.
This does not make metabolic targeting futile. It means the unit of treatment must sometimes be the relationship rather than the individual pathway. Blocking a transporter may prevent exchange between cells. Altering fibroblast activation may cut off a nutrient source. Normalizing blood vessels may improve oxygenation and reduce selection for aggressive hypoxic states. Combining metabolic therapy with immunotherapy may restore both nutrient availability and immune function.
The practical goal is not to make the tumor completely nutrient-free, which is impossible within a living person. It is to disrupt the specific exchanges and adaptations that the tumor depends upon more than normal tissue does.
The Tumor Microenvironment Is a Metabolic Battleground
An activated T cell is itself metabolically demanding. To multiply and attack, it must take up glucose and amino acids, increase glycolysis, maintain mitochondrial function, synthesize membranes, and produce cytokines. Natural killer cells, dendritic cells, macrophages, and other immune cells also require specific metabolic programs.
Cancer cells compete for many of the same resources. A rapidly growing tumor can consume glucose, glutamine, arginine, tryptophan, oxygen, and other nutrients faster than they are delivered. It can release lactate, adenosine, kynurenine, potassium, reactive oxygen species, and lipids that alter immune-cell behavior. Poorly formed blood vessels intensify the problem by producing regions of hypoxia and acidosis.
T cells entering this environment may be asked to perform under conditions resembling exhaustion and starvation. Low glucose can restrict the glycolytic burst needed for attack. Low oxygen changes gene expression through hypoxia-inducible factors. Lactate and acidity impair cytokine production and movement. Adenosine signals through inhibitory receptors. Tryptophan breakdown through IDO and related enzymes produces kynurenine, which can suppress effector cells and favor regulatory T cells.
The result is not simply a weak immune system. It is a locally engineered metabolic environment that changes what immune cells are able to do. This helps explain why a patient may have abundant tumor-infiltrating lymphocytes that nevertheless fail to control the cancer.
Metabolic intervention may therefore strengthen immunotherapy in several ways. It can reduce lactate export, block adenosine signaling, preserve amino acids, improve mitochondrial fitness, or normalize blood vessels. It can also backfire. A drug that restricts a nutrient needed by cancer may restrict the same nutrient in T cells. The challenge is to identify differences in timing, transporters, enzymes, and flexibility that allow tumor metabolism to be targeted while preserving immune metabolism.
Macrophages Can Be Metabolically Re-educated
Macrophages are highly adaptable. In one setting they engulf pathogens and produce inflammatory signals. In another they repair tissue, promote blood-vessel growth, and quiet inflammation. Tumors often exploit the repair program. Lactate, hypoxia, lipids, cytokines, and dying cells can encourage macrophages to support angiogenesis, matrix remodeling, and immune suppression.
This is not a simple division between “good” and “bad” macrophages. Tumor-associated macrophages occupy a range of states and can change over time. Their metabolism helps establish those states. Therapies that alter lactate, fatty-acid oxidation, iron handling, or mitochondrial function may change macrophage behavior even when they do not directly kill cancer cells.
Mitochondria Can Move Between Cells
One of the most striking recent developments is the recognition that mitochondria are not always confined to the cell in which they were produced. Cells can transfer mitochondria through tunneling nanotubes, extracellular vesicles, partial cell fusion, or other forms of contact. In damaged tissues, transfer of healthy mitochondria may help rescue stressed cells. Tumors can exploit the same process.
A 2025 Nature study found mitochondrial DNA mutations in tumor-infiltrating lymphocytes that matched mutations in nearby cancer cells. The researchers showed that cancer-cell mitochondria could enter T cells. These transferred mitochondria resisted normal disposal through mitophagy and contributed to mitochondrial dysfunction, senescence, and impaired immune activity. Tumor mitochondrial mutations were also associated with poorer responses to checkpoint inhibitors in the patient groups examined].
A complementary 2026 Cell Metabolism study described transfer in the opposite direction. Tumor cells acquired mitochondria from several types of immune cells. The immune cells lost functional capacity, while the acquired mitochondria fused with the cancer cells’ mitochondrial network. Mitochondrial DNA leakage then activated cGAS-STING and type I interferon programs that favored immune evasion and lymph-node metastasis in the experimental models.
These findings are still new, and their importance will vary among tumors. They should not yet be treated as a universal explanation for metastasis or immunotherapy failure. Nevertheless, they expand the metabolic ecosystem in a profound way. Cells may exchange not only glucose, lactate, or amino acids, but entire energy-producing organelles.
Mitochondrial transfer also complicates the language of “healthy” and “damaged” mitochondria. A cancer cell with impaired respiration may acquire mitochondria that improve oxidative capacity. A T cell may receive tumor-derived mitochondria that carry mutations or inhibitory signals. Mitochondrial quantity, origin, quality, and compatibility may all influence the result.
Future therapies might attempt to block the structures used for transfer, restore mitophagy in recipient immune cells, prevent the fusion of acquired mitochondria, or selectively target cancer cells that become dependent on transferred organelles. These strategies remain experimental, but the underlying biology is an important addition to the metabolic theory of cancer.
Genes and Metabolism Form a Feedback Loop
The genetic and metabolic theories of cancer are often presented as rivals, but the biology repeatedly shows that they are interdependent. Oncogenic mutations reprogram metabolism. Metabolic conditions influence gene expression, DNA repair, chromatin structure, and the selection of new mutations. Each can become both cause and consequence.
PI3K-AKT-mTOR signaling increases glucose uptake, protein synthesis, lipid synthesis, and nutrient use. MYC increases glycolysis, glutamine metabolism, nucleotide synthesis, and mitochondrial production. Loss of p53 can reduce metabolic restraint, alter antioxidant responses, and increase glycolysis. RAS mutations can stimulate glucose uptake and macropinocytosis. Hypoxia stabilizes HIF proteins, which increase glycolytic enzymes, lactate transport, angiogenic signaling, and survival pathways.
Metabolism can in turn alter genetic regulation. Acetyl-CoA supplies acetyl groups used in histone acetylation. S-adenosylmethionine supplies methyl groups for DNA and histone methylation. Alpha-ketoglutarate is required by enzymes that remove methyl groups. Succinate, fumarate, and 2-hydroxyglutarate can inhibit those enzymes. NAD+ affects sirtuins and DNA-repair enzymes. Lactate can contribute to histone lactylation.
These relationships mean that nutrient availability can influence which cellular programs are active. A chronic metabolic state may help lock a cell into an abnormal identity. Conversely, a mutation in a metabolic enzyme can generate a metabolite that reorganizes the epigenetic landscape.
IDH Mutations: When Genetics Creates an Oncometabolite
Mutations in isocitrate dehydrogenase, or IDH, provide one of the clearest examples. Normal IDH enzymes participate in cellular metabolism. Mutant IDH1 or IDH2 acquires a new activity and produces large amounts of D-2-hydroxyglutarate. This molecule resembles alpha-ketoglutarate and interferes with enzymes involved in DNA and histone demethylation. The result is an abnormal epigenetic state that impairs normal differentiation.
Here, a genetic mutation creates a metabolic product, and that metabolic product changes gene regulation and cell identity. The categories cannot be separated.
The clinical importance became clear in the INDIGO trial. Vorasidenib, a brain-penetrant inhibitor of mutant IDH1 and IDH2, significantly prolonged progression-free survival and delayed the need for another intervention in patients with grade 2 IDH-mutant glioma [. This is metabolic oncology at its most precise: identify a mutation-defined enzyme, identify the oncometabolite it produces, and block that abnormal activity in the patients whose tumors carry the dependency.
The success of mutant-IDH inhibition does not validate every metabolic therapy. It demonstrates the value of specificity. The strongest metabolic targets may be those in which the tumor has created an unusual metabolite, enzyme activity, transporter, or redox requirement that normal tissues do not share to the same degree.
Treatment Creates New Metabolic States
Cancer treatment does more than reduce the number of cells. It changes the environment and therefore changes evolution within the tumor. Oxygenation may improve when tumor mass shrinks, or worsen when blood vessels are damaged. Dead cells release nutrients. Inflammation increases. Surviving cells face oxidative stress, DNA damage, and nutrient disruption. These pressures select for metabolic traits that were previously less important.
Drug-tolerant persister cells illustrate this process. These cells survive concentrations of therapy that kill most of the tumor, often without a permanent resistance mutation. They may enter a slow-cycling state, use autophagy to recycle components, improve mitochondrial quality control, increase fatty-acid oxidation, or strengthen antioxidant systems. If treatment is removed, some can return to growth. During their period of survival, they may acquire stable genetic resistance.
Several studies have found increased oxidative phosphorylation in persister populations. Others have identified dependence on glutathione, aldehyde metabolism, lipid handling, or iron. There is no single persister metabolism across all cancers, but redox control is a recurring theme.
This creates an opportunity. Cells that survive one therapy may expose a second vulnerability. A persister cell using oxidative phosphorylation may become sensitive to mitochondrial inhibition. A cell heavily dependent on GPX4 to prevent lipid peroxidation may become vulnerable to ferroptosis. A cell using autophagy for survival may be affected by disrupting lysosomal recycling.
In a foundational study, therapy-resistant, mesenchymal-like cancer cells showed dependence on the lipid-peroxide repair enzyme GPX4. Inhibiting GPX4 induced ferroptotic death in these otherwise resistant cells. This led to the broader concept that the metabolic state selected by treatment may contain its own Achilles’ heel.
The difficulty is timing. A vulnerability may exist only during a temporary state. Treating too early, too late, or continuously may miss it or damage normal cells. Sequential therapy - one treatment to force the tumor into a constrained state, followed by another to target that state - may be more effective than giving every agent simultaneously.
Ferroptosis and the Management of Oxidative Risk
Ferroptosis is an iron-dependent form of cell death caused by uncontrolled oxidation of membrane lipids. Cells prevent it through systems involving glutathione, GPX4, cystine uptake through system xC-, coenzyme Q, and lipid-remodeling enzymes. Cancer cells with high oxidative activity may depend heavily on these protections.
Ferroptosis connects several metabolic themes: iron, fatty-acid composition, cysteine and glutamine use, NADPH, mitochondrial activity, and antioxidant defense. It also shows why antioxidants cannot be discussed in a simple good-or-bad framework. Antioxidant systems protect normal tissue, but cancer cells may use the same systems to survive treatment. Conversely, indiscriminately increasing oxidation can injure normal tissue and may not reach the tumor in the needed form.
The therapeutic goal is selective loss of redox control in vulnerable cancer cells. This may involve blocking cystine uptake, depleting glutathione, inhibiting GPX4, changing membrane lipid composition, or combining ferroptosis pressure with radiation, targeted therapy, or immunotherapy. Most such strategies remain investigational, but they represent one of the most active areas of metabolic oncology.
What the Ecosystem Model Means for Metabolic Therapy
The metabolic approach to cancer remains promising, but the newer evidence changes how that promise should be described. The early vision often imagined that cancer had one universal metabolic defect and could therefore be starved through a broadly applied intervention. The current vision is more precise: different tumors contain different metabolic populations, and those populations change in response to treatment.
A useful metabolic target should meet several conditions. The tumor should depend on it strongly. Normal tissues should be less dependent or better able to compensate. The pathway should remain important in the patient’s actual tumor environment. Alternative fuels and stromal support should be limited. The drug or intervention should reach the relevant cells. Biomarkers should identify the people most likely to benefit.
Mutant IDH meets many of these conditions. Some other targets have not yet done so. Telaglenastat, an inhibitor of glutaminase, had a strong preclinical rationale and encouraging early findings. Yet in the randomized CANTATA trial, adding telaglenastat to cabozantinib did not improve progression-free survival in metastatic clear-cell renal-cell carcinoma. The trial did not prove that glutamine metabolism is unimportant. It showed that inhibiting one point in that pathway was not sufficient in an unselected clinical setting.
Metformin provides another caution. Laboratory studies, epidemiology, and its effects on insulin and mitochondrial complex I generated great interest. However, the large MA.32 randomized trial found that adding metformin to standard treatment did not improve invasive disease-free survival in patients with high-risk operable breast cancer without diabetes. Metformin may still be useful for diabetes, insulin resistance, or selected tumor contexts, but it should not be presented as a generally proven anticancer metabolic drug.
These results are not failures of the metabolic perspective. They are evidence that metabolism must be treated with the same precision expected in molecular oncology. A drug does not succeed merely because it affects a pathway cancer uses. The relevant tumor must be dependent on that pathway at the time the drug is given.
Ketogenic Diets, Fasting, and Fuel Restriction
Ketogenic diets lower carbohydrate intake, reduce insulin in many people, increase fatty-acid oxidation, and raise circulating ketone bodies. Fasting and fasting-mimicking approaches can reduce glucose, insulin, IGF-1, and amino-acid signaling while activating stress-response and recycling pathways in normal tissues. These changes have plausible anticancer effects and have produced encouraging results in many laboratory models.
Human evidence remains more limited and heterogeneous. Small studies and case series suggest that ketogenic approaches can sometimes be implemented during treatment and may improve selected metabolic measures, quality of life, or treatment tolerance. A 2025 clinical report in glioblastoma described implementation of dietary ketogenic metabolic therapy, but the study design does not establish that the diet itself caused improved survival. Larger controlled studies are still needed to determine which cancers, stages, treatment combinations, and dietary forms are most likely to benefit.
The ecosystem model explains why results may vary. Some tumors cannot use ketone bodies effectively; others can. Some become more dependent on fatty-acid oxidation. The type of dietary fat can change membrane composition, inflammation, and ferroptosis susceptibility. A diet that lowers glucose may also alter immune cells, the microbiome, body weight, hormones, and drug metabolism. In an underweight patient, excessive restriction may reduce treatment tolerance. In an insulin-resistant patient with excess adiposity, improving metabolic health may provide benefits even if the tumor is not directly starved.
It is therefore more accurate to view diet as a systemic metabolic intervention rather than a simple fuel switch. Its effects depend on the person, the tumor, and the treatment. Dietary strategies are most scientifically defensible when they are nutritionally adequate, coordinated with treatment, monitored with meaningful biomarkers, and evaluated for their actual goals rather than assumed to work through a universal mechanism.
Combination Therapy and Adaptive Pressure
Metabolic therapies may be most useful in combinations that prevent escape. A glycolysis inhibitor may be paired with a mitochondrial strategy. Glutamine pressure may be combined with inhibition of compensatory glucose use or antioxidant defense. Lactate transport inhibition may be paired with checkpoint blockade. Ferroptosis-inducing strategies may be used after targeted therapy creates a persister state.
The sequence matters. Restricting nutrients before chemotherapy may slow proliferation and reduce sensitivity to drugs that target dividing cells, yet the same intervention might increase oxidative stress and improve the effects of radiation. A short metabolic intervention may protect normal cells during treatment, while prolonged restriction may impair immune recovery. The answers will differ among cancers and therapies.
The future is likely to involve adaptive treatment designs that measure the tumor’s response and change the intervention accordingly. Rather than prescribing one permanent metabolic program, clinicians may use time-limited metabolic pressure to create a vulnerability and then exploit it.
The Whole Body Is Part of the Tumor’s Environment
Cancer metabolism does not end at the edge of the tumor. The liver regulates glucose, ketones, amino acids, and lipids. Muscle stores and releases amino acids and lactate. Adipose tissue supplies fatty acids and produces hormones and inflammatory signals. The gut microbiome modifies nutrients, bile acids, and immune activity. The nervous and endocrine systems coordinate stress responses. Cancer can alter all of these systems.
Advanced tumors may cause insulin resistance, increased liver glucose production, muscle breakdown, altered fat metabolism, and cachexia. The tumor can draw resources from the entire body while inflammatory signals change how normal tissues use nutrients. In this setting, aggressive dietary restriction may not starve the tumor selectively. It may accelerate loss of muscle and reduce resilience.
At earlier stages, systemic metabolic health can still matter. Hyperinsulinemia activates growth signaling. Excess adipose tissue can increase estrogen production, inflammatory cytokines, and altered adipokines. Fatty liver and metabolic syndrome change circulating nutrients and immune tone. Physical activity improves insulin sensitivity, muscle metabolism, vascular function, and immune mobilization. Sleep and circadian disruption affect glucose regulation, hormones, and immune function.
These influences should not be reduced to blame or to the claim that lifestyle alone causes or cures cancer. Tumors arise through many pathways, and metabolically healthy people can develop aggressive disease. The practical point is that the tumor exists within a physiological environment. Improving that environment may reduce supportive signals, increase treatment tolerance, preserve muscle, and improve overall health even when it does not directly eliminate cancer.
The whole-body perspective also argues for better endpoints in metabolic research. A therapy may lower tumor glucose uptake but worsen muscle wasting. Another may not shrink the tumor immediately but may improve immune function or treatment tolerance. Meaningful evaluation requires attention to tumor control, survival, body composition, symptoms, metabolic markers, and quality of life.
How Metabolic Oncology Is Becoming More Precise
The next phase of metabolic oncology depends on measurement. It is not enough to know that cancer, in general, uses glucose or glutamine. Clinicians need to know what a particular tumor is using, which cells are using it, and whether that dependency remains present after treatment begins.
Metabolomics measures many small molecules in tissue or blood. Stable isotope tracing follows labeled nutrients into downstream products. Hyperpolarized magnetic resonance imaging can visualize selected metabolic reactions in real time. PET tracers are being developed for nutrients and pathways beyond glucose. Spatial transcriptomics and proteomics show where enzymes and transporters are expressed. Single-cell methods reveal distinct metabolic states within the same tumor.
Each method has limitations. Gene expression does not prove metabolic flux. A metabolite level may rise because production increased or because consumption decreased. A biopsy is spatially limited. Imaging has finite resolution. Blood signals are influenced by the whole body. The strongest assessment will combine several kinds of evidence.
Biomarkers will be essential for clinical trials. A glutaminase inhibitor should ideally be tested in tumors with demonstrated glutamine dependence and limited bypass capacity. An MCT inhibitor should be matched to tumors with active lactate exchange. A ferroptosis strategy may require evidence of GPX4 dependence, iron availability, and vulnerable membrane lipids. Dietary studies should document glucose, insulin, ketones, body composition, nutrient intake, tumor genotype, and concurrent therapy rather than treating every ketogenic diet as equivalent.
Organoids and patient-derived models may also help. A tumor sample can be grown in three-dimensional culture with selected stromal or immune components, then exposed to different nutrient conditions and drugs. These systems will never reproduce the entire person, but they may identify vulnerabilities more accurately than standard cell lines.
Ultimately, precision metabolic oncology may resemble infectious-disease treatment more than a universal diet. The clinician identifies the organism’s vulnerabilities, considers the environment in which it lives, anticipates escape routes, and uses combinations or sequences that reduce the chance of resistance.
A Metabolic Map Must Be Updated Over Time
A tumor’s metabolic profile at diagnosis may not remain valid after chemotherapy, hormone therapy, immunotherapy, radiation, or metastasis. A prostate cancer deprived of androgen may increase its reliance on other substrates. A targeted therapy may select oxidative persister cells. A metastasis in the liver may encounter different nutrients from a metastasis in bone or brain.
Repeated assessment is therefore likely to matter. Liquid biopsies, metabolic imaging, blood markers, and occasional tissue sampling may help show whether the tumor has changed. The goal is not to create an impossibly detailed map of every pathway. It is to identify the few dependencies that have become difficult for the tumor to escape.
What This Perspective Adds to the Metabolic Theory of Cancer
The metabolic theory made an essential contribution by challenging the idea that mutations alone explain cancer. It restored attention to mitochondria, fermentation, nutrient use, and the physiological environment in which cancer grows. It also encouraged researchers and patients to ask whether the metabolic differences between normal and malignant cells could be exploited therapeutically.
The ecosystem perspective does not discard that contribution. It refines it in several ways.
First, metabolic dysfunction does not have to mean uniform respiratory failure. A cancer may contain cells with defective respiration, cells with vigorous oxidative phosphorylation, and cells that alternate between the two. Mitochondria can be abnormal yet indispensable.
Second, the Warburg effect is not simply a mechanism for making ATP. It supports biosynthesis, redox balance, rapid adaptation, acid production, lactate exchange, and signaling.
Third, cancer metabolism is not confined to cancer cells. Fibroblasts, immune cells, adipocytes, endothelial cells, the extracellular matrix, and distant organs participate in the supply and handling of nutrients.
Fourth, metabolites are not merely fuels. They regulate enzymes, gene expression, chromatin, differentiation, inflammation, and immune behavior.
Fifth, treatment changes metabolism. Surviving cells can enter temporary states with new dependencies. These states may be therapeutically exploitable, but only if they are identified and targeted at the correct time.
Finally, metabolic therapy is unlikely to be one universal method of starving cancer. Its strongest future lies in identifying constrained dependencies, disrupting resource exchange, restoring immune metabolism, and combining metabolic pressure with standard treatment in a personalized and adaptive way.
This is a more complex view than the original Warburg model, but it is also more biologically realistic. It explains why a tumor can be strongly glycolytic and still depend on mitochondria, why reducing glucose may slow one tumor but not another, why stromal cells matter, why immunotherapy is affected by lactate and nutrient competition, and why treatment resistance can emerge without an immediate new mutation.
Promising Directions for the Next Several Years
Several areas appear especially likely to shape the field.
Spatial metabolic mapping will reveal how nutrient use differs across tumor regions and how those regions change during treatment. Stable isotope studies in patients will distinguish metabolism that occurs in human tumors from metabolism that appears mainly in laboratory media.
Immunometabolism will continue to connect metabolic therapy with checkpoint inhibitors, cellular therapies, vaccines, and innate immunity. Strategies that reduce lactate, adenosine, or nutrient competition may help immune cells recover function. Equally important will be protecting T-cell mitochondria and redox balance.
Mitochondrial transfer has opened an unexpected field of research. Investigators will need to determine how common it is in human cancer, which cell types participate, how it affects treatment, and whether it can be blocked safely.
Ferroptosis and other forms of oxidative cell death will remain major targets because treatment-resistant cells often depend on sophisticated lipid and antioxidant defenses. The challenge will be to create selectivity and to understand how immune cells and normal tissues respond.
Oncometabolites will likely produce additional precision therapies. Mutant IDH is the clearest example, but tumors with fumarate hydratase deficiency, succinate dehydrogenase deficiency, altered methionine metabolism, or unusual nucleotide dependencies may offer other opportunities.
Diet-drug interactions will receive more rigorous study. Rather than asking whether a ketogenic diet or fasting works for cancer in general, trials can ask whether a defined metabolic intervention improves the effect of a specific therapy in a biomarker-selected population. They can also compare dietary compositions, timing, and effects on body composition and immunity.
Finally, mathematical and computational models may help predict adaptation. A tumor ecosystem contains feedback loops and competing populations. Models that combine genomics, spatial data, metabolomics, and treatment response could help identify combinations that close the most important escape routes without creating excessive toxicity.
Otto Warburg was correct that altered metabolism is central to cancer. The major refinement is that there is no single metabolic abnormality shared in exactly the same way by every cancer cell. Tumors organize multiple metabolic strategies across different cells, regions, and stages of disease.
Some cells ferment glucose and export lactate. Others import that lactate and oxidize it. Some depend on glutamine, fatty acids, or amino acids supplied by surrounding tissue. Fibroblasts and adipocytes become nutrient partners. Immune cells are placed under metabolic pressure. Metabolites alter gene expression. Mitochondria are remodeled, selectively removed, or even transferred between cells. Treatment kills sensitive populations while creating new metabolic states in the survivors.
Seen in this way, cancer is not merely a cell with a defective power plant. It is a living metabolic ecosystem capable of reorganizing its supply lines, sharing resources, changing its priorities, and adapting to attack.
That complexity should not be mistaken for hopelessness. Complex systems often contain vulnerabilities that simpler models miss. A cooperative exchange can be interrupted. A metabolite-producing mutation can be blocked. A persister state can create dependence on mitochondrial respiration or antioxidant defense. Immune cells can be metabolically restored. A tumor that can switch between two fuels may become vulnerable when both the preferred pathway and the escape pathway are addressed.
The future of metabolic oncology will probably not rest on a single diet, drug, or universal theory. It will rest on learning how each tumor is organized, which resources it cannot do without, which cells provide those resources, how treatment changes the system, and when a metabolic intervention can be applied with the greatest selectivity.
The Warburg effect remains the doorway into this field. The metabolic ecosystem is the larger room it has opened.