The Warburg effect is ubiquitous in cancer. It consists of the upregulation of glucose uptake, glycolysis, and subsequent lactate secretion, sometimes by over 200 times, in cancer cells as compared to normal cells. A common explanation for this phenomenon is that tumor cells can be oxygen deprived, because they are far away from the diffusive range of oxygen in the blood supply. But if that's the case, then how are they close enough to blood supply to have access to enough glucose? How can glucose diffuse to places where oxygen cannot, even if oxygen is a much smaller molecule?

  • 2
    $\begingroup$ This is a very good question. I think this is in fact a major problem for the whole glycolysis/hypoxia theory. And I'm not aware of any evidence for rapid glycolysis in hypoxic tumor regions. But note that Warburg's finding was that tumor cells have high glycolysis also in the presence of oxygen; that is completely different from glycolysis in hypoxia. $\endgroup$
    – Roland
    Commented Jan 30, 2016 at 15:48
  • $\begingroup$ This is a great systems bio question. You could also take it a step back from cancer and ask something like "what are the absolute fluxes (input - output) of O2 and glucose in healthy tissue, and how does one affect the other?" $\endgroup$
    – tel
    Commented Jan 30, 2016 at 22:35

3 Answers 3


I think this is a difficult question for which the answer is not clear --- to my knowledge there is little data on metabolism in solid tumors, and no clear consensus in the scientific community. So I'm not going to attempt to give a definitive answer, but here's at least a few thoughts / opinions.

Why transformed cells engage in aerobic glycolysis, or the "Warburg effect", is a question people have been trying to answer for nearly a hundred years now. There are dozens of hypotheses, but none of them is broadly accepted. One popular hypothesis laid out by Gilles & Gatenby says that transformed cells from solid tumors are glycolytic because they have been selected for anaerobic metabolism in a hypoxic environment, and this selection is somehow irreversible, so that the cells stay glycolytic even after being extracted from the tumor and cultured in aerobic conditions. I think this hypothesis is the origin of the idea of hypoxic and yet glycolytic tumors.

In a way the Gilles & Gatenby hypothesis is a clever idea --- it rests on the well-accepted fact that most cells become glycolytic in hypoxia, and rhymes with the concept of selection for survival in tumors. But the problem with the argument (as you have realized) is that the tumor environment would have to be low in oxygen and at the same time high in glucose. I agree with you that it is unlikely that such an environment could arise due to lack of blood supply, since oxygen diffuses more easily through tissues (and cells) than glucose does. Moreover, considering the high demand for glucose by glycolytic cells, glucose should become limiting long before oxygen runs out. So one can argue that poorly vascularized tumors should be mainly oxidative, not glycolytic!

To my knowledge, there is no data demonstrating existence of (regions of) solid tumors with low oxygen and high glucose. There is of course ample evidence that many solid tumors consume large amounts of glucose, and also many studies showing hypoxic regions in tumors. But most likely these are different types of tumors, or different regions. In an environment with neither oxygen nor glucose, cells will simply die --- which is exactly what happens in the necrotic cores of larger solid tumors.

The best source of further information is probably the tumor physiology literature. This review is a good starting point.

Finally, another issue with Gilles & Gatenby's hypothesis is that aerobic glycolysis can easily be demonstrated in cells that have never been in a tumor. For example, in vitro transformation can increase glycolysis, and normal lymphocytes induce rapid glycolysis when activated by receptor stimulation. So there is no need to assume that glycolytic tumors are hypoxic.

  • $\begingroup$ Personally I think aerobic glycolysis is poorly worded (though I know it is common). in my opinion ahypoxic hyperglycolysis describes the phenotype better :) $\endgroup$
    – jiggunjer
    Commented Mar 16, 2016 at 6:34

The outside of the tumor has access to the most nutrients and oxygen, sometimes even the blood supply (angiogenesis). Solid tumors have this microenvironment if you will, and the human tumor microenvironment often suffers from a lack of oxygen and nutrients, and acidosis (Milosevic et al., 2004). The distance from the outside of the tumor to the core has an established diffusion gradient of oxygen, nutrients, etc. (Cristini et al., 2005, especially since blood vessels can't always penetrate to these areas). A take home point is the larger the tumor becomes, the less nutrients available to the inward tumor cells until the cells often become necrotic or necroptotic (Huang et al., 2013). Furthermore, since tumors maintain a constant inward pressure, the dead cells are essentially eaten by the living cells and everything is pushed to the center, thereby providing the interior environment of the tumor materials to work with (Greenspan, 1972).

There are gross metabolic shifts explained below that provide the tumors different ways of obtaining nutrients. I don't think it has so much to do with glucose diffusing as deep into the tumor as oxygen as it does this metabolic reprogramming.

How does the tumor alter it's expression profile to fit it's needs under the stress of starvation? Example based approach

Hypoxia-inducible factors or HIF unfortunately modulate tumor progression. (See Bellot et al., 2009, Zhang et al., 2008 for examples of autophagy induced by HIF, Huang et al., 2014 for a non-exhaustive overview of additional ways HIF modulates cancer progression). Tumor cells suffering chronic hypoxia upregulate their GLUT transporters and glycolytic pathways via action from HIF-1α/ß because they must often begin to produce ATP without oxygen. They also, however, downregulate mitochondrial activity through transactivation of PDK1 and MXI1 Denzo et al., 2008. PDK1 ends up inhibiting PDH, and MXI1 ends up actually reducing the number of mitochondria per cell.

The incorporation, then, of glutamine into the TCA cycle where pyruvate would normally be needed supplements the now-missing carbon requirement (see figure below) where pyruvate would be involved, because referring again to the Denzo paper, in the hypoxic tumor pyruvate is converted into lactate by LDH, in part explaining the acidosis. Incorporation of glutamine however should purportedly be independent of HIF expression, so I looked at papers such as Ma et al., 2013, which showed that starving tumors of glucose ended up making them more aggressive due largely to regulation by PKCζ, and subsequent incorporation of glutamine metabolism.

enter image description here

Fig A. Looking here at alternative fuel pathways during starvation of the tumor, the cells can scavenge from the necrotic center of the tumor, utilize autophagy, other carbon source, etc. to fit their needs. Source: White, E., 2013

More specifically, there is a paradigm shift in tumor cells deficient in PKCζ that involves (1) upregulation of Phosphoglycerate dehydrogenase (PHGDH) and Phosphoserine aminotransferase (PSAT1), which are normally repressed by PKCζ, (2) uninhibited mTOR activity, which is normally shut down during nutrient depletion, and (3) inhibited AMPK and autophagy pathways, which are normally active in nutrient depletion.

"Genome-wide transcriptome analysis demonstrated that PKCζ-deficient cells displayed gene alterations in pathways consistent with the use of glutamine as a glucose alternative."

In both humans and mice they conclude that with respect to intestinal cancers, PKCζ regularly acts as a tumor suppressor. But the major point here is that multiple pathologic mutations can be implicated here to try and make sense of the Warburg effect.

Leithner et al., 2015 likewise shows that tumor cells (this is specific for lung cancer) can begin to utilize lactate to form glucose building blocks by modifying the gluconeogenesis pathway, mediated largely by a mitochondrial isoform of the phosphoenolpyruvate carboxykinase (PEPCK) known as PCK2:

"Under low glucose, all three carbons from (13)C₃-lactate appeared in the PEP pool, further supporting a conversion of lactate to pyruvate, via pyruvate carboxylase to oxaloacetate, and via PCK2 to phosphoenolpyruvate."

I'd like to bring up again that in the hypoxic tumor, evidence supports that pyruvate is converted to lactate, but now I imply the lactate can potentially be converted to glucose, a positive feedback loop providing the tumor glucose. This is somewhat supported by the clinical study in Koukourakis et al., 2006 that shows tumors overexpressing LDH-5 and HIF-1α/2α that in the malignant tumors, oxygen consumption was low and glucose consumption was high. But I also want to go back to the Denzo paper because they brought up some interesting points, and I recommend reading it: The decreased mitochondrial need for oxygen might be more important than the increased glycolysis.

Some interesting points:

1) The metabolic shift induced by hypoxia lowers the tumor oxygen requirement and heightens it's glucose requirement. This doesn't necessarily mean the tumor is actually getting glucose, and sometimes, it doesn't mean the tumor cell is actually deprived of oxygen, either, rather the hypoxia is induced by simply not enough oxygen to fit the high oxygen requirement, read Takakusagi et al., 2014.

2) By reprogramming it's metabolism and cannibalizing cells that actually did die from deprivation, the tumor can viably "get" energy -though not necessarily directly in the form of glucose- through a number of pathological mutations.

3) Anaerobic glycolysis followed by LDH activity is an efficient adaptation if you look at Koppenol, Bounds & Dang, 2011, the hypoxic tumor cell can produce 26 ATP and 26 lactic acid from 13 glucose in the time it takes oxidative phosp. to produce 36 ATP from one glucose. In a time of starvation efficient energy production is a good thing, and the hypoxic mechanism is fast, especially if we consider the tumor is producing glucose from lactic acid in any capacity. They also point out that tumors become in a way addicted to the Warburg effect,

"... oncogenic deregulation of biomass accumulation for cell proliferation creates an increased, sustained bioenergetic demand that addicts cancer cells to an adequate anabolic supply."

"How can glucose diffuse to places where oxygen cannot, even if oxygen is a much smaller molecule?" I don't have such a definitive answer for this, though I would personally determine that it doesn't, not by much. Where I want to go is it's not really about glucose and oxygen getting to the same place. I suspect that they reach just about equally as deep into the tumor depending on the composition of the tumor itself and the tissue it resides in. I also want to point out that hypoxia is not the same as anoxia, where there is no oxygen and the cells are for the most part dead in those regions. If you read the length of this post, you understand that the tumor cells are experiencing hypoxia, or a deficient amount of oxygen reaching them, and likewise I would assume a deficient amount of glucose, as opposed to a lack thereof. In that sense again, and because there's not a lot of nutrients to go around, we're alluding to the efficient usage of both glucose and oxygen, and beneficial biomass accumulation, which we demonstrated the Warburg effect is key to. Because once you get to the point of anoxia, again, we're looking at anoxic cell death, but those cells get turned into what essentially amounts to recycling material.

Closing Notes

I don't see why glucose wouldn't be able to diffuse along it's gradient into the tumor considering there is a microfluid environment inside the tumor, but where we're talking about in the hypoxic region (not the anoxic regions) I suspect there is still some glucose present just as there's some oxygen. It's important, then, for the tumor cells to have more of the GLUT transporters from the action of HIF, and the efficient usage of whatever metabolites it can get. I generally lack data for how far into a tumor they can actually perfuse, but to me it would be hard to quantify since there are many confounding factors such as what tissue the tumor originated from, irregularities in the microcirculation, etc.

  • 1
    $\begingroup$ I don’t think any of this answers the question. $\endgroup$
    – Roland
    Commented Jan 30, 2016 at 16:21
  • $\begingroup$ Thanks for the useful information, but I agree with Roland, this does not answer my question. $\endgroup$
    – a06e
    Commented Jan 30, 2016 at 18:39
  • $\begingroup$ I appreciate your answer. I think the most relevant aspect is that you think somehow there can be a high enough concentration of glucose in places where oxygen is not accessible, thus creating a niche for enhanced glycolysis. This is what I find tough to digest, because oxygen is so small a molecule, it should be very easy for it to diffuse, thus it should be available wherever glucose is available. But if you can find some references supporting your claim that would be great, since that is the essence of my question. $\endgroup$
    – a06e
    Commented Jan 30, 2016 at 22:09
  • $\begingroup$ The link to the Milosevic paper is broken. I think you copy/pasted a link that goes through your university proxy or something, which I can't access. Can you fix it please? $\endgroup$
    – a06e
    Commented Jan 30, 2016 at 22:12
  • $\begingroup$ Do you have a reference with the diffusion coefficients of glucose and oxygen? $\endgroup$
    – a06e
    Commented Feb 1, 2016 at 18:43

Hypoxia does not mean no oxygen reaches the place, it means low oxygen tension when compared to normal well perfused tissue. In my experiments I have observed the exterior of rat glioma models well perfused but not so well inside. We have used MR based methods to study perfusion and IHC to study vasculature. In the same way, I have observed increased MCT-4 expression (indirectly, more lactic acid production) in the outer regions of the tumor but certainly sometimes there are some pockets inside with increased MCT-4. But no matter what, the whole of the tumor and surroundings is acidic. This could be due to lactic acid or maintenance of intracellular pH by proton exchangers. Though there is no clear answer, there are lots of points on which I agree with CMosychuk. Low oxygen tension increases HIF activity which has its repercussions. I also faced another curious situation. I was working with pseudohypoxic tumors. They don't have low oxygen tension but still have activated HIF because HIF is artificially stabilized thanks to some mutations (VHL, SDHB). Now, I expected more GLUT because I could see increased 18F-FDG uptake in these tumors but I saw increased hexokinase-2 expression. Remember, in this case there was no hypoxia, still activation of HIF had its effect that is increased glycolysis! So I too think that low oxygen tension might mean low nutrient supply but it certainly does not mean no nutrient supply. We see tumor parts having low oxygen tension, happily living and metabolizing and making the whole region acidic. I have not even seen a necrotic core in these tumors. When, I killed these cells giving temozolomide to the animals, the acidity was not there anymore. So, acidity was certainly from metabolism of cells. Remember, these are all in vivo studies and we do different kinds of measurements on the same tumor as the rat is alive.

  • 1
    $\begingroup$ Hello Jyotsna, welcome to Biology.SE and thank you for your contribution to the site. That you posted such an extensive response is great, however it would be very much appreciated if you could list some of the sources for your claims. $\endgroup$
    – Ebbinghaus
    Commented Mar 15, 2016 at 18:59
  • $\begingroup$ You're right that hypoxia (and hypoglycemia) is a matter of degrees, of course. But then you would need to argue from quantitative data. What does "well perfused" mean? What is the oxygen tension in your tumors? Respiration rate? Glucose concentration, glycolytic rate? These values are typically very difficult to determine in vivo. $\endgroup$
    – Roland
    Commented Mar 15, 2016 at 20:08

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .