by Stephen Western

Dietary changes are an appealing way to adapt to the reality of a cancer diagnosis, as diet is one aspect of healing that is completely within the patient’s control. A fact well-known in the field of oncology is that every tumour and tumour type has a unique genetic profile, with an individual set of mutations, copy number alterations, protein expression, and so forth. Consequently different tumours, and especially tumours originating in different tissues, or different subtypes of tumours arising in the same tissue (for example glioblastoma versus IDH1-mutant astrocytoma) may have different metabolic requirements and different sensitivities to various forms of nutrient deprivation. Just as there is no one anti-cancer drug that works equally well for each type of cancer, there is likely no single anti-cancer diet. Ideally, a diet should be tailored to the unique sensitivities of one's tumour type, if such knowledge exists.

What to Include

There is an extensive literature on foods with potential anti-cancer properties. Diet is such a matter of personal taste that it’s probably better for each individual to do much of their own research on dietary choices. The bulk of anti-cancer foods fall under the vegetable, fruit, or spice headings. A few basic guidelines could be:

  • Maximize micronutrient dense, low calorie vegetable foods

  • Overall fruit consumption should be monitored due to the sugar content, though many kinds of berries are notable for their antioxidant, anti-cancer properties

  • Determine how much protein your body actually requires (probably somewhere in the neighborhood of 8-10% of calories as protein (0.65-0.85 grams protein per kilogram body weight) and limit protein intake to that level. Requirement for certain amino acids will go up however, when recovering from surgery, or radiochemotherapy

  • Try to keep a stable blood glucose level by avoiding too much sugar, refined starch and high glycemic-index foods

  • Limit omega-6 fatty acids (mostly oils derived from seeds) and include the long chain omega-3 oils EPA and DHA, found in cold-water fish. The importance of choosing the right fats deserves its own section.

Fats and Oils

The main thing to remember when choosing fats and oils in your diet is that cancer thrives on omega-6 fatty acids, in particular the 20-carbon arachidonic acid. Cancer cells use arachidonic acid as the precursor for building inflammatory prostaglandins (like PGE2) and leukotrienes, which are cancer promoting and a major causative factor for brain inflammation. Some of the arachidonic acid in our cells is assembled from smaller chain omega-6 fatty acids, such as linoleic acid. However, a much larger proportion may be consumed directly in foods. Eggs are inherently high in arachidonic acid, because it is required for embryo development. Grain-fed meats may also be a plentiful source of dietary arachidonic acid. Many seed oils are high in the omega-6 linoleic acid, with the exception of flax oil and a few others.

Another way to limit cancer-promoting inflammation is to include plentiful amounts of long-chain omega-3 fatty acids such as EPA and DHA, which compete for the same enzymes as omega-6 oils and have an anti-inflammatory action. The major source of EPA and DHA is various cold-water fish species. It is important to make sure that your fish oil product has been tested for mercury and other industrial contaminants. The optimal dose of fish oil hasn’t been firmly established in clinical trials, though a daily dose of 3 grams of combined EPA and DHA was shown to have benefit in a trial of metastasized (secondary) brain tumour patients.

Specific nutrients and nutraceuticals (food-derived chemicals with drug-like properties) will be discussed in the Supplements section.

Nutritional Solutions, founded by Jeanne Wallace, works with cancer patients to design cancer-fighting diets, and has extensive experience with brain tumour patients.

Strategies for limiting tumour access to nutrients

For further ideas, see also Targeting Tumour Metabolism


A study with glioma-bearing rats tells us that depending on context, dietary iron supplementation could either be harmful or beneficial to cancer patients (18). Iron is a critical nutrient for cell proliferation and high-grade gliomas have elevated expression of transferrin receptors (19). Transferrin is the main iron carrier protein. In the rat study (18), male Wistar rats were implanted into the hip with a chemically-induced rat glioma cell line (strain 35), and after formation of a tumor node, the rats were divided into several treatment groups. Control rats were given tap water containing 0.2-0.3 mg/L of iron ions (Fe2+). Another group of rats was given drinking water supplemented with 60-63 mg/L iron ions (Fe2+). In the iron-supplemented group, tumours grew faster than in the control group, and iron-supplemented rats had 34% shorter lifespans than the control group. In marked contrast, when rats were given a single 15 Gy dose of radiation to the tumour area, rats supplemented with iron had a much better response to radiation and longer survival than rats treated with radiation but no iron supplementation. These irradiated, iron-supplemented rats lived twice as long as the untreated control rats. Glioma cells in the irradiated, iron-supplemented group died by a combination of apoptosis and ferroptosis, a form of iron-dependent cell death. Thus, iron appears as an effective radiosensitizer, but can actually stimulate tumour growth outside of the context of radiation.

Caloric restriction

Multiple experiments with mice have shown the potential benefit of caloric restriction (CR) in significantly slowing the growth of implanted brain tumours. These experiments used the highly angiogenic mouse astrocytoma CT-2A, implanted into immunocompetent mice. This model may be considered to mimic highly angiogenic human gliomas, particularly glioblastomas.

In one of these studies (8), the mice were put on a diet of regular mouse chow, calorically restricted by 30% compared to the non-restricted control group. This degree of caloric restriction caused the mice to lose 30% of their body weight over a two week period, at which time all the animals were sacrificed.

The results of this caloric restriction on tumour growth and biology are quite impressive. By the end of two weeks, tumour weight was reduced by 65% in the CR group compared to control mice. Blood glucose was reduced by 63% and ketone bodies were elevated by 114% in the CR mice.

Figure 2 Mulrooney 2011 Caloric restriction

Effects of caloric restriction (CR) versus ad libitum (AL) feeding in glioma-bearing mice

On a molecular level, activation of the transcription factor nuclear factor kappa beta (NF-KB) (see discussion in Targeting Invasion) was significantly reduced, as were levels of the COX-2 enzyme (see discussion of Celebrex in Repurposed Drugs). Both of these molecules are implicated in malignant glioma biology and contribute to processes such as invasion, angiogenesis, and immunosuppression. The macrophage-attracting chemokine CXCL2 (or MIP-2), and the macrophage marker CD68 were significantly reduced in the CR mice, evidence of decreased macrophage infiltration into the tumour (see Targeting Tumour-Associated Macrophages/Microglia).

Similar studies of caloric restriction in mice with experimental gliomas detail additional benefits and mechanisms by which caloric restriction can hinder tumour growth and invasiveness. One caveat is that the metabolism of mice is roughly seven times faster than human metabolism. Therefore a more severe caloric restriction would be required to replicate these results in humans. The degree of weight loss is a more precise indicator of the effects of caloric restriction. As the mice in this study lost 30% of body weight over two weeks, this degree of caloric restriction is clearly rather extreme, and would not be sustainable in the long-term, though long-term maintenance of a body weight in the low-normal range would likely be beneficial.

The stimulating effects of caloric restriction on anti-cancer immunity is explored on the Re-educating the Immune System page.

Protein restriction and IGF-1

If you live in one of the more affluent countries of the world, you likely take in more protein than your body actually requires. Of the three calorie sources (carbs, protein, and fat), protein probably has the best reputation as being good for health, so we don't often hear about the benefits of reducing protein intake.

The average intake for North Americans and northern Europeans is around 15-18% of total calories as protein. This is significantly more than the recommended daily allowance of 10% of calories as protein (or about 0.83 grams protein per kilogram body weight). The RDA is designed to meet the protein requirement of 97.5% of the adult population. The median protein requirement for a healthy adult is even less than this: 0.65 grams of protein per kilogram body weight, or perhaps 8% of total calories (9). Okinawans, a population with very low rates of cancer, take in about 9% of their calories as protein (10).

Insulin-like growth factor 1 (IGF-1), as its name implies, promotes cellular growth. By attaching to its receptor (IGF1R), it initiates activation of the infamous PI3K/Akt/mTOR signaling pathway, which is overactive in many cancers, including both high- and low-grade astrocytomas (11). In fact, one of the main drugs currently being tested in clinical trials for low grade glioma is everolimus, an mTOR inhibitor. This is where the importance of protein restriction comes in.

Protein restriction versus caloric restriction

Both animal and human studies have found that protein intake is the major determinant of serum IGF-1 levels. One study (9) put a group of humans on a 20% caloric restriction diet for a year. While this moderate long-term caloric restriction reduced insulin levels and C-reactive protein (an inflammation marker), serum IGF-1 was not reduced in this group. The investigators then compared members of the Calorie Restriction Society (consuming on average 1800 kcal per day, 24% of total calories from protein) with non calorically restricted vegans eating only 10% of total calories as protein. Both groups had similar low levels of insulin and C-reactive protein, but the low-protein vegans had lower levels of total and free IGF-1, despite their higher body weight, body mass index, and body fat content. To further prove the correlation of low protein intake to reduced IGF-1, six members of the caloric restriction group reduced their protein intake by 43%, to a level just above that of the average low-protein vegans, for three weeks. This intervention led to a 25% decrease in serum IGF-1.

The connection between protein intake, IGF-1 levels, and tumour growth has been demonstrated in controlled animal studies. The same group which published the previously described study (9) also conducted a study comparing 7% to 21% of total calories as protein in mice engrafted with prostate and breast cancer cells (10). When the mice were primed with their 21% or 7% protein diets for four weeks prior to prostate cancer cell implantation, the difference in protein consumption led to a 81% reduction in tumour weight for the lower protein group by the end of the study. When protein restriction was initiated four weeks after tumour cell implantation, the lower protein group had 50% lighter tumours than the higher protein group by the end of the study. Significantly, there was no difference in body weight, or glucose levels between the two groups, showing that glucose levels had no impact on the tumour growth restriction seen in this study. Impressively, the prostate cancer xenografts barely grew in the protein restricted mice.

In the same study, the investigators then compared protein restriction to chemotherapy with the mTOR inhibitor everolimus, as well as protein restriction and everolimus combined. The protein restricted mice (7% of total calories as protein) and the mice given everolimus had a similar degree of tumour growth inhibition. The tumours were smallest when protein restriction was combined with everolimus. To explain this effective tumour inhibition, levels of phosphorylated (active) mTOR, phosphorylated S6K (a downstream target of mTOR) and Ki67 (a marker of cell proliferation) were tested. Protein restriction highly significantly reduced each of these three markers, as did everolimus therapy alone, with the greatest inhibition following combined protein restriction and everolimus. The authors note that protein restriction is more powerful than calorie or fat restriction in lowering IGF-1 levels.

Along the same lines, in a separate study (12), mice were fed various macronutrient defined diets for nine days, then tested for changes in blood glucose and IGF-1. Amazingly, mice on a "ketogenic" diet (90% of calories from fat, 9% of calories from protein, and only 1% of calories from carbohydrates) in unlimited quantities had no significant change in glucose or IGF-1, showing the futility of trying to reduce glucose levels by carb restriction alone, without caloric restriction. Other groups of mice were fed low protein diets (4% of total calories), and no protein diets, among other diets, for 9 days. Another group was fasted for 60 hours and used as a reference. Of all the diets tested, only protein restriction or protein elimination significantly reduced IGF-1 levels. None of the diets significantly reduced glucose levels. The fasted mice (60 hour water-only) had highly significant reductions in both glucose and IGF-1. This experiment shows that in mice, it is difficult to lower glucose levels without caloric restriction or fasting. Attempting to restrict carbs without caloric restriction may be futile in terms of reducing blood glucose. When mice were given various diets with a 50% caloric restriction, there was still very little change in glucose levels. Only the 60 hour fast could highly significantly reduce glucose levels. The authors also studied the effect of fasting and calorie restriction combined with chemotherapy. In every case, fasting prior to chemotherapy protected healthy cells from the treatment while sensitizing cancer cells. This encouraging result will be described in another section.

These studies showing the efficacy of protein restriction in reducing IGF-1 levels and slowing tumour growth are very encouraging, for while caloric restriction may provide benefits in the short-term, consuming fewer calories than the body requires cannot be sustainable in the long-term. The benefits of protein restriction, on the other hand, are independent of caloric restriction, and may therefore be sustained in the long-term.

P53-mutant tumours and glucose deprivation

Most TP53 gene mutations in cancer cause the overexpression of mutant p53 proteins, which have aquired oncogenic gain-of-function. It is not simply a matter of loss of p53 tumour suppressor function, but a gain of tumour-promoter function by the mutant proteins. TP53 mutations are nearly universal in IDH-mutant astrocytomas, and are also common in IDH-mutant oligoastrocytomas and certain glioblastoma subtypes. A study (16) published in 2012 in the journal Cell Cycle by a group from Georgetown University in Washington DC provides evidence that mutant p53 protein accumulation can be inhibited by a severely carbohydrate restricted diet, leading to significantly inhibited tumour growth in mice with mutant-p53 tumours.

The researchers first showed that the expression of mutant p53 proteins can be inhibited in vitro by depleting the glucose concentration in the cell medium. Conversely, mutant p53 expression was not affected by serum or amino acid depletion. This inhibition of mutant p53 accumulation following glucose depletion was found to be due to protein degradation (breakdown) via an autophagic process rather than by proteasome activity.

This breakdown of mutant p53 by glucose restriction was then tested in mice. p53-mutant (p53 A135V) transgenic mice and wild-type (non-mutant) mice were randomized to one of three diets: a normal mouse chow, a low-carbohydrate diet, and a high-carbohydrate diet. The low carbohydrate diet was 74% protein, 24% fat and 2% carbs by caloric content, or 71.7% protein, 10% fat, and 1.9% carbohydrates by weight (the remainder being fiber, vitamins and minerals). By caloric content, the standard diet and high-carb diets were 53% and 69% carbohydrate, 20% and 17% protein.

Blood tests showed that mice on the low-carb diet had significantly reduced fasting blood glucose levels: around 100 or <100 mg/dl versus about 130 mg/dl in the standard chow group. After four months on the various diets, the mice were sacrificed. Remarkably, mutant p53 levels were reduced in the mammary glands, ovaries and fat of the p53-mutant, low-carb diet group. Conversely, wild-type p53 was stabilized in the wild-type p53, low-carb diet group.

Further testing was done to determine the effects of carbohydrate restriction on tumour growth. Mice were fed either the standard diet or the low-carb diet for two weeks, then implanted with mouse mammary cancer cells which contained either wild-type p53, mutant p53 (G242A), or lacking p53. Three to four weeks later, tumour volumes were assessed. On the standard diet, tumours were largest in the p53-mutant group, followed by p53 null. Tumours were smallest in the p53 wild-type group. The low-carb diet effectively inhibited tumours in all three groups, but most significantly in the p53-mutant and p53-null groups. The extracted tumours were analyzed for p53 levels. In agreement with the in vitro studies, the low-carb diet stabilized p53 in the wild-type group, but inhibited mutant p53 in the mutant group.

This study is in agreement with a previous study by a different group which showed that colon cancer xenografts in which p53 was knocked out were inhibited by metformin treatment, while the same xenografts with normal p53 expression were completely resistant to metformin treatment (17). Metformin is a diabetic drug which lowers and stabilizes blood glucose levels.

This evidence raises questions about the relative efficacy of protein restriction versus carbohydrate restriction for p53 mutant cancers. A different study showed that reducing the percentage of calories from protein from 21% to 7% can inhibit tumour growth in breast and prostate xenografts (10), while the study described above showed tumour inhibition with a high protein, very low-carb diet. I could find no data on the p53 status of the cell lines used in the low-protein study. Future investigations studying the effects of dietary restrictions on tumour growth should note the p53 status of the cell lines, and compare the effects of protein restriction versus carbohydrate restriction in p53-mutant tumours. There are many different p53 mutations found in cancer, about half of them found in one of about 6 "hotspots" on the TP53 gene. In the above study, 6 different p53 mutants were all inhibited by glucose restriction in vitro.

Conclusions: While I wouldn't go so far as to recommend a high-protein diet (see below), the evidence in this study is strongly in favour of a low-glycemic, carbohydrate-restricted diet especially for tumours bearing mutant p53. As stated above, mutant p53 is found in approximately 95% of IDH-mutant astrocytomas.

Ketogenic Diet

Trials are currently underway testing the ketogenic diet for cancer patients. The diet, originally designed for epileptics, consists of liberal fats and oils, with tightly restricted protein and carbohydrate intake, with the aim of reducing circulating glucose (and by extension, insulin) levels to a low-normal range. After several weeks on this diet, the body re-tools to using fatty acids and ketone bodies (ketones) as an alternative fuel source replacing glucose. Even the brain, when fully adapted, can function on ketones as roughly two-thirds of its fuel supply. The theory is that cancer cells are not able to metabolize ketones (which are fatty acid breakdown products) as normal cells can (2).

However, this ability may vary depending on the tumour type. The enzyme encoded by the gene OXCT1 is the rate-limiting enzyme involved in ketone body catabolism. As reported in a recent study (3), 10 out of 17 glioblastomas were negative for this enzyme, while zero out of 5 anaplastic astrocytomas were negative. The implication is that the ketogenic diet may work more effectively in glioblastomas which lack ketolytic enzymes, though this remains to be proven in a clinical trial.

A major drawback of the ketogenic diet is its restrictiveness and high fat content. Some find it unappealing and ultimately unsustainable. In studies with mice, the ketogenic diet in unrestricted amounts did not slow experimental brain tumour growth (4). In these studies, the ketogenic diet slowed tumour growth only when given in restricted quantities which caused the mice to begin losing weight. Overall caloric restriction is likely a more effective way to slow tumour growth than a re-distribution of caloric sources. In one case report of a glioblastoma patient on a severely calorically restricted ketogenic regimen (600kcal/day), the diet appeared to be effective for a time (5). However, a person cannot continue a calorically restricted diet causing weight loss indefinitely. My advice would be to cut consumption of sugars, grains and starches down to a low level, choosing low glycemic-index foods, in order to keep blood glucose at a stable level, while maintaining a body weight in the low range of normal.

In some cases, a non-calorically restricted ketogenic diet may be effective, as it was for two pediatric astrocytoma cases (6) and for four (one minor response, three stabilizations) of 18 recurrent glioblastoma patients assessed in the ERGO trial (7). More recently, a small study (13) in Pennsylvania investigated six newly diagnosed glioblastoma patients who adopted a ketogenic diet in combination with standard treatment. While the study was too small to determine efficacy, median time-to-progression/recurrence was 10.3 months. In four evaluable patients, mean non-fasting blood glucose on the ketogenic diet during radiation therapy was 84 mg/dl, down from 142.5 mg/dl prior to starting the diet and one week before starting radiation therapy. This drop in blood glucose was statistically significant. Only one or two of the six patients were intentionally restricting calories.

The ERGO (14) trial was a pilot trial of a non-calorically restricted ketogenic diet carried out in Germany. The trial design consisted of 20 glioblastoma patients, with a median of 2 recurrences per patient. 3 patients dropped out due to a lowered quality of life on the diet. Almost all the patients achieved ketosis, as measured in urine samples. Of the 17 evaluable patients, 3 had disease stabilization for 11-13 weeks (2.5-3 months) and one patient had a minor response. Total response plus stabilization rate was therefore 24%. The six-month progression-free survival rate was 0%, and the median PFS was 1.1 months. After progression on the ketogenic diet, 7 patients received salvage therapy with bevacizumab.

The results of the ERGO trial are best compared with the phase III trial of NovoTTF versus chemotherapy for recurrent glioblastoma (15), as this trial was carried out at about the same time, and patients in that trial also had a median of 2 recurrences before trial start. In this phase III trial, median PFS was 2.1-2.2 months (chemo vs TTF) and six month progression-free survival rate was 15.1-21.4% (chemo vs TTF). The authors of the ERGO trial conclude that a non-calorically restricted ketogenic diet "has no significant clinical activity when used as single agent in recurrent glioma". A follow-up trial (ERGO2), testing a ketogenic diet and transient fasting during re-irradiation for recurrent glioblastoma, is currently recruiting patients in Frankfurt, Germany. This combination approach including caloric restriction is far more likely to be effective, and a ketogenic diet combined with radiation led to a cure rate of over 80% in the GL261 glioma mouse model, as described on the Standard of Care page.

The following trials are also currently testing ketogenic diets for glioblastoma: NCT01865162, NCT02046187, NCT01535911

For details on pharmacological methods of lowering blood glucose, see the discussion of metformin in the Repurposed Drugs section, and berberine in the Supplements section.


  1. Metabolic reprogramming: a cancer hallmark even Warburg did not anticipate. Ward et al. 2012.

  2. Targeting energy metabolism in brain cancer: review and hypothesis. Seyfried et al. 2005.

  3. Ketolytic and glycolytic enzymatic expression profiles in malignant gliomas: implication for ketogenic diet therapy. Chang et al. 2013.

  4. The calorically restricted ketogenic diet, an effective alternative therapy for malignant brain cancer. Zhou et al. 2007.

  5. Metabolic management of glioblastoma multiforme using standard therapy together with a restricted ketogenic diet: case report. Zuccoli et al. 2010.

  6. Effects of a ketogenic diet on tumor metabolism and nutritional status in pediatric oncology patients: two case reports. Nebeling et al. 1995.

  7. The ERGO trial: A pilot study of a ketogenic diet in patients with recurrent glioblastoma. Rieger et al. 2010 ASCO annual meeting abstract.

  8. Influence of caloric restriction on constitutive expression of NF-KB in an experimental mouse astrocytoma. Mulrooney et al. 2011.

  9. Long-term effects of calorie or protein restriction on serum IGF-1 and IGFBP-3 concentration in humans. Fontana et al. 2008.

  10. Dietary protein restriction inhibits tumor growth in human xenograft models. Fontana et al. 2013.

  11. Activation of PI3K/mTOR pathway occurs in most adult low-grade gliomas and predicts patient survival. McBride et al. 2010.

  12. Dietary- and fasting-based interventions as novel approaches to improve the efficacy of cancer treatment. Sebastian Brandhorst, January 2013. Dissertation.

  13. Targeting metabolism with a ketogenic diet during the treatment of glioblastoma multiforme. Champ et al. 2014.

  14. ERGO: A pilot study of ketogenic diet in recurrent glioblastoma. Rieger et al. 2014.

  15. NovoTTF-100A versus physician’s choice chemotherapy in recurrent glioblastoma: A randomised phase III trial of a novel treatment modality. Stupp et al. 2012.

  16. Dietary downregulation of mutant p53 levels via glucose restriction: mechanisms and implications for tumor therapy. Rodriguez et al. 2012.

  17. Systemic treatment with the antidiabetic drug metformin selectively impairs p53-deficient tumor cell growth. Buzzai et al. 2007.

  18. Effects of Iron Ions and Iron Chelation on the Efficiency of Experimental Radiotherapy of Animals with Gliomas. Ivanov et al. 2015.
    READ ABSTRACT Email me for a PDF copy

  19. Transferrin receptors and glioblastoma multiforme: Current findings and potential for treatment. Voth et al. 2015.
    READ ABSTRACT Email me for a PDF copy

This page was created on 01/26/2014 and last updated on 04/21/2019

Our privacy / cookie policy has changed.
Click HERE to read it!