Protein, Ketogenesis, and Glucose Oxidation

In our last post, we discussed the relationship between protein and blood sugar in ketogenic dieters.

Despite all the evidence we have brought to bear suggesting that increased protein does not increase GNG, there is an important line of argument that does support the idea that increased protein increases GNG. Although the data is indirect, and some of it is poorly documented, it is compelling.

This supporting argument is the relationship between protein and glucose oxidation (the use of glucose as fuel). As we mentioned in our last post, the rate of GNG is not what we really care about as keto dieters.

What we want to know is whether excess protein leads to using a higher total amount of glucose as fuel. The amount of glucose oxidation matters, because the benefits we expect to gain from a keto diet are probably a result of using ketones for fuel instead of glucose whenever we can.

In Brief

  • Excess protein probably results in lower ketone levels. Although there is a paucity of hard clinical evidence, there are several reasons to believe this is true.
  • There appears to be an inverse relationship between ketone levels and glucose oxidation.
  • Therefore, increasing protein probably increases glucose oxidation.
If so, then
  • Eating more protein would reduce the benefits of a ketogenic diet, by making it less ketogenic, and increasing glucose oxidation.
  • This would need to be reconciled with the combined evidence that I/G appears to be the main determinant of ketogenesis, and yet doesn't appear to change in keto dieters eating protein.
As always, we'd like to see experimental confirmation of these ideas instead of just relying on the "chains of plausible mechanism" outlined here. Chains of plausible mechanism can be broken by a single weak link, or by other effects that we didn't take into account.

Protein and Ketogenesis

It is already widely believed and asserted that excess protein reduces ketone production. For example, this is stated in Peter Attia's blog, which we highly recommend , and in Volek and Phinney's excellent book, The Art and Science of Low Carbohydrate Living ¹. However, as far as we are aware, there aren't any experiments that measure this directly and without confounders. The mechanism cited is the rise in insulin that protein induces. In our previous article, we presented evidence that the insulin-to-glucagon ratio is not significantly changed in response to protein in ketogenic dieters. That article also cites evidence that the insulin-to-glucagon ratio (I/G) is an accurate predictor (and perhaps even cause) of glucose regulation. Moreover, there is evidence that ketogenesis is itself regulated by the insulin-to-glucagon ratio ²³. So we don't find that explanation particularly compelling. Nonetheless, there may be other lines of evidence that we are not yet aware of.

The following indirect argument suggests that protein inhibits ketogenesis. There appears to be an inverse relationship between ketosis and blood sugar . We have already shown that protein raises blood sugar in ketogenic dieters. Together, this would seem to indicate that protein decreases ketosis.

If protein inhibits ketogenesis, then the following argument can be made that protein increases glucose oxidation. It would make intuitive sense that higher blood ketone concentrations would correspond to lower levels of glucose oxidation, since ketones can usually replace glucose for fuel. In fact, in some studies, an inverse relationship has been shown to hold between glucose oxidation and serum ketone levels in people fasting for short periods , and in epileptic children . (See also the the Randle Cycle.) Therefore, if protein inhibits ketogenesis, it very likely increases glucose oxidation.

Can we determine the effect of protein on glucose oxidation directly?

Scientists do have ways to measure glucose oxidation, for example through indirect calorimetry. We can measure respiratory quotient (RQ): the proportion of oxygen in exhalation is used to infer the proportion of fat and glucose being used, by taking advantage of the fact that oxidizing fat and oxidizing glucose require different amounts of oxygen. Then you can combine this with resting energy expenditure (REE), a measure of calories expended, to determine the total amount of glucose being used for energy. Observations that would indicate more glucose oxidation include: higher energy expenditure at the same RQ, or higher RQ at the same energy expenditure.

If such an effect is confirmed, knowing its magnitude would be equally important. I.e., how much extra glucose oxidation would be expected from a certain amount of excess protein? Is it linear, or is there a large effect at the beginning, and very little effect after, or some other relationship? All of this is far from clear to us. We would love to see it addressed experimentally, since even though we are inclined to believe it, there are some potential confounders, including changes in fat and calorie intake, and differential effects of different types of fat and protein.


There is evidence that protein does not increase the rate of GNG. There is evidence that I/G, which appears to control glucose production and ketogenesis, does not change in keto dieters when they eat protein. Nonetheless, there are compelling arguments that protein increases glucose oxidation in keto dieters. Experiments will need to be done to reconcile these seeming contradictions.

In the meantime, limiting protein to levels that are known to be adequate seems prudent.

Further Reading

Lucas Tafur has an interesting and relevant article entitled Safe starches, blood glucose and insulin.


Evidence type: authority

Both insulin and glucose (probably by causing the secretion of insulin) suppress ketones. This is why, for example, consuming more than about 50 gm of carbohydrates per day and/or more than about 120-150 gm of protein per day makes it difficult to be in nutritional ketosis – too much insulin secretion.

¹ Evidence type: authority
Stephen J. Phinney and Jeff S. Volek. The Art and Science of Low Carbohydrate Living: An Expert Guide to Making the Life-Saving Benefits of Carbohydrate Restriction Sustainable and Enjoyable. Beyond Obesity LLC (May 19, 2011). ISBN-10: 0983490708

Another reason to avoid eating too much protein is that it has a modest insulin stimulating effect that reduces ketone production. While this effect is much less gram-for-gram than carbohydrates, higher protein intakes reduce one's keto-adaptation and thus the metabolic benefits of the diet.

² Evidence type: review of experiments
Foster DW, McGarry JD. The regulation of ketogenesis. Ciba Found Symp. 1982;87:120-31.

(Emphasis ours)


Ketone bodies accumulate in the plasma in conditions of fasting and uncontrolled diabetes. The initiating event is a change in the molar ratio of glucagon:insulin. Insulin deficiency triggers the lipolytic process in adipose tissue with the result that free fatty acids pass into the plasma for uptake by liver and other tissues. Glucagon appears to be the primary hormone involved in the induction of fatty acid oxidation and ketogenesis in the liver. It acts by acutely dropping hepatic malonyl-CoA concentrations as a consequence of inhibitory effects exerted in the glycolytic pathway and on acetyl-CoA carboxylase (EC The fall in malonyl-CoA concentration activates carnitine acyltransferase I (EC such that long-chain fatty acids can be transported through the inner mitochondrial membrane to the enzymes of fatty acid oxidation and ketogenesis. The latter are high-capacity systems assuring that fatty acids entering the mitochondria are rapidly oxidized to ketone bodies. Thus, the rate-controlling step for ketogenesis is carnitine acyltransferase I. Administration of food after a fast, or of insulin to the diabetic subject, reduces plasma free fatty acid concentrations, increases the liver concentration of malonyl-CoA, inhibits carnitine acyltransferase I and reverses the ketogenic process.

³ Evidence type: experiment (non-human animals)
Ubukata E, Mokuda O, Sakamoto Y, Shimizu N. Effect of various glucagon/insulin molar ratios on blood ketone body levels in rats by use of osmotic minipumps. Diabetes Res Clin Pract. 1996 Sep;34(1):1-6.


The bihormonal control by insulin and glucagon of blood ketone body level was studied. Mixed solutions with various molar ratios of glucagon and insulin (G/I) were subcutaneously infused continuously for five days by use of the osmotic minipump in the normal rats. The concentrations of insulin and glucagon solution were set at the high G/I molar ratio, the moderate G/I molar ratio and the low G/I molar ratio. In addition, the moderate G/I molar ratio group was divided into three sub-groups: low glucagon and low insulin, moderate glucagon and moderate insulin, and high glucagon and high insulin. After five days, the rats were decapitated to measure plasma ketone body, free fatty acid (FFA), glucose, insulin and glucagon. The FFA level was not significantly different among three groups. The glucose level was not different between the high and moderate G/I molar ratio groups, and decreased in the low G/I molar ratio group. 3-beta-hydroxybutyrate (3-OHBA) and acetoacetate (AcAc) levels in the high G/I molar ratio group were elevated, and 3-OHBA level in the low G/I molar ratio group was lowered compared to those in the moderate G/I molar ratio group. Among three moderate G/I molar ratio sub-groups, there was no difference in 3-OHBA and AcAc levels. These results demonstrate that plasma ketone body levels are controlled by the plasma G/I molar ratio.

Evidence type:
Fukao T, Lopaschuk GD, Mitchell GA. Pathways and control of ketone body metabolism: on the fringe of lipid biochemistry. Prostaglandins Leukot Essent Fatty Acids. 2004 Mar;70(3):243-51.

(Emphasis ours)


Ketone bodies become major body fuels during fasting and consumption of a high-fat, low-carbohydrate (ketogenic) diet. Hyperketonemia is associated with potential health benefits. Ketone body synthesis (ketogenesis) is the last recognizable step of lipid energy metabolism, a pathway that links dietary lipids and adipose triglycerides to the Krebs cycle and respiratory chain and has three highly regulated control points: (1) adipocyte lipolysis, (2) mitochondrial fatty acids entry, controlled by the inhibition of carnitine palmityl transferase I by malonyl coenzyme A (CoA) and (3) mitochondrial 3-hydroxy-3-methylglutaryl CoA synthase, which catalyzes the irreversible first step of ketone body synthesis. Each step is suppressed by an elevated circulating insulin level or insulin/glucagon ratio. The utilization of ketone bodies (ketolysis) also determines circulating ketone body levels. Consideration of ketone body metabolism reveals the mechanisms underlying the extreme fragility of dietary ketosis to carbohydrate intake and highlights areas for further study.

(But note that this suggests that there may be a pathway for suppression by elevated insulin alone.) Evidence type: review of experiments
Amanda E. Greene, Mariana T. Todorova, Richard McGowan, Thomas N. Seyfried. Caloric Restriction Inhibits Seizure Susceptibility in Epileptic EL Mice by Reducing Blood Glucose. Epilepsia. Volume 42, Issue 11, pages 1371–1378, November 2001. DOI: 10.1046/j.1528-1157.2001.17601.x
Our findings in mice together with those in humans indicate that CR, like fasting, lowers blood glucose levels while inducing ketosis (1,41–44). This contrasts with studies of the KD, in which blood glucose levels are not reduced in association with ketosis (5,15). It is interesting to note that the antiseizure effect of the KD was greater when it was administered under restricted than under ad libitum conditions (12), suggesting that reduced blood glucose levels may enhance the efficacy of the KD. Despite evidence for an inverse relation between blood glucose and ketone levels in normal humans and humans with epilepsy under fasting or the KD (45), little attention has been given to the possibility that these dietary therapies prevent seizures through an effect on blood glucose levels. From previous neurochemical studies and from our statistical analyses, we show that blood glucose levels determine both blood ketone levels and seizure susceptibility in EL mice and emphasize the importance of blood glucose as a predictor of epileptogenesis in this epilepsy model.

Evidence type: experiment
J.A. Romijn, M.H. Godfried, M.J.T. Hommes, E. Endert, H.P. Sauerwein. Decreased glucose oxidation during short-term starvation. Metabolism, Volume 39, Issue 5, May 1990, Pages 525–530. http://dx.doi.org/10.1016/0026-0495(90)90012-2

(Emphasis ours)


Prolonged fasting (for days or weeks) decreases glucose production and oxidation. The effects of short-term starvation (ie, < 24 hours) on glucose metabolism are not known. To evaluate this issue, glucose oxidation and glucose turnover were measured after 16-hour and subsequently after 22-hour fasting. Glucose oxidation was calculated by indirect calorimetry in 12 healthy men (age 22 to 44 years); glucose turnover was measured by primed, continuous infusion of 3-3H-glucose in eight of these 12 volunteers. After 16-hour fasting net glucose oxidation was 0.59 ± 0.17 mg · kg−1 · min−1 and glucose tissue uptake 2.34 ± 0.12 mg · kg−1 · min−1. No correlation was found between net glucose oxidation and glucose tissue uptake. Prolonging fasting with an addtional 6 hours resulted in decreases of respiratory quotient (0.77 ± 0.01 v 0.72 ± 0.01) (P < .005), plasma glucose concentration (4.7 ± 0.1 v 4.6 ± 0.1 mmol/L) (P < .05), glucose tissue uptake (2.10 ± 0.12 mg · kg−1 · min−1)(P < .05), net glucose oxidation (0.09 ± 0.04 mg · kg−1 · min−1)(P < .005), and plasma insulin concentration (8 ± 1 v 6 ± 1 mU/L) (P < .005). Net glucose oxidation expressed as a percentage of glucose tissue uptake decreased from 22% ± 8% to 2% ± 1% (P < .05). There was no net glucose oxidation in seven of 12 controls after 22-hour fasting. Serum free fatty acid (FFA) concentration (364 ± 34 to 575 ± 48 μmol/L) (P < .005) and plasma ketone body concentration (104 ± 23 to 242 ± 38 μmol/L) (P < .005) increased between 16- and 22-hour fasting. After 16-hour fasting an inverse correlation was found between ketone body concentration and net glucose oxidation (P < .05) and between ketone body concentration and net glucose oxidation expressed as a percentage of glucose tissue uptake (P = .07). No significant correlation could be demonstrated between FFA and ketone body concentration and between FFA and net glucose oxidation. It is concluded that glucose oxidation decreases rapidly even within 1 day of starvation. This may be explained by physiological mechanisms like decreased insulin action and/or inhibition of glucose oxidation by ketone bodies, even in relatively low concentrations.

Evidence type: experiment
M. W. Haymond, C. Howard, E. Ben-Galim, and D. C. DeVivo. Effects of ketosis on glucose flux in children and adults AJP - Endo October 1, 1983 vol. 245 no. 4 E373-E378


Sequential glucose flux [rate of appearance - rate of disappearance] studies were carried out in five normal and six epileptic children and ten adult volunteers using [6,6-2H2]glucose to determine the effect of ketosis on carbohydrate homeostasis in children and adults. All subjects were studied after 14 and 30-38 h of fasting while consuming a normal diet and the epileptic children under 14 h of fasting while consuming an isocaloric ketogenic diet (75% fat wt/wt). Glucose flux, when expressed per kilogram body weight, was inversely correlated with the degree of ketosis in children (P less than 0.001) and in adults (P less than 0.01), but not when both children and adults were considered together (r = 0.078). When glucose flux was corrected for estimated brain weight, the relationship between glucose flux and ketonemia was linearly related in children (P less than 0.001), in adults (P less than 0.02), and when all subjects were considered together (P less than 0.001). The inverse relationship between ketonemia and glucose flux corrected for estimated brain mass is consistent with a partial replacement of glucose by ketone bodies for cerebral metabolism and may provide a more rational means of expressing glucose flux data to take into account the higher brain-to-body ratio in children.


An inverse relationship was observed between ketone body concentration and glucose utilization whether expressed on a body weight or estimated brain weight basis. The relationship between glucose utilization (on a brain weight basis) and ketone body concentration may not be completely linear because glucose utilization appears to approach a minimum of 20-30 µmol·min⁻¹ 100 g estimated brain⁻¹ at ketone body concentration of 5 mM or greater. This observation suggests that a basal requirement for glucose utilization may exist that cannot be supplanted by ketone bodies regardless of their plasma concentration and is in keeping with the observation that a basal glucose oxidation rate is required by brain tissue for optimum utilization of ketone bodies (13).


  1. Nice post. Thanks.

    This protein++ => glucose++ appears here and there, both experimentally and also in practice.

    For instance Rosedale likes to keep protein in middle ranges just because of that effect. Ketogenic diets for cancer treatment often do not produce benefits unless CR is involved. The reason may very well be lower protein (glucose is already low, and fat comes from internal stores as needed). Protein also rises insulin, with different amino-acids having different potential but even if lower then carbs, its still enough to prevent first step in ketogenesis. This is nicely analyzed on calories proper blog (highly recommended read).

    How to turn this into practical point. Since this is yet another restriction I doubt it would motivate people.

    Supplements of course. GSE is intriguing. There are number of papers out there and its work on all 3 steps in ketogenesis ( here is one from head). Niacin is also interesting... niacin inhibits HSL which reduces lipolysis. This later leads to rebound effect and lipolysis is increased for some time which I experienced myself. This fact is used in detoxing procedure, for instance see Dr. Yu Niacin detox protocol and its effect on lipid oxidation are not controversial.

    So, its good to know the theory. But without supplements, there isn't much you could do about it.

    1. Hi! I agree that many people with clinical experience already assert the protein/glucose connection, and I'm sure they have reasons to believe it. They may even have relevant data in some form. I hope they decide to make it available.

      I also think you are exactly right that there are important possible confounders that have to be separated, not least of which is calories vs. protein.

      I'm not at all against supplements, but I'm also interested in doing the best we can without them. To me the critical part of whether a restriction is motivating or do-able, depends on the interplay of how much it could benefit you, how likely it is to actually give you the benefits, and how practical it is. (Cf. How to Judge a Health Practice).

      In particular, in this case, this means to me that it doesn't leave you feeling hungry. I believe hunger is a warning signal from the body that shouldn't be ignored. This is one of the greatest differences between calorie restriction and ketogenic diets, which share mechanisms. So I think that if you restrict protein so much that it makes you physically uncomfortable, that is not sustainable, and it's probably not good for you either.

  2. Thank you very much for an interesting and thought-provoking series of posts. The following thoughts occurred to me as a result:

    1. Surely the question "does excess protein inhibit ketogenesis?" can be easily answered by measuring blood ketone levels multiple times before, during and after protein consumption? Blood ketone levels, not blood glucose levels, are what indicate if one is in ketosis or not.

    2. If the digestion of protein causes cells to slow their uptake of glucose in order to improve their uptake of amino acids*, then the rate of glucose removal from the blood will be reduced, causing blood glucose levels to rise**. If this is the case one would expect the demand for gluconeogenesis to decrease in response, not increase, as it is after all a costly metabolic process.

    * Do you know if there is an inverse relationship between glucose and amino acid uptake rates in cells at a fixed insulin/glucagon ratio?
    ** With reference to the table in your post entitled, "Protein, Gluconeogenesis, and Blood Sugar" (20 Jan 2013), the level that blood sugar rose to in individuals on a ketogenic diet is still very low. Lower, in fact, than the fasting glucose levels of test subjects on the glycolytic diet.

    1. Hi Gustav, thank you for reading. Those are good points.

      1. Yes, I don't think it would be hard. I would even think enough relevant data exists somewhere. Many anecdotes support it, including my own.

      In the post about blood sugar, I outlined some experiments one could do on oneself to find out the amount and duration of response, and one could do the same with ketones. The main problem is expense. Three weeks of experiments involving 10 readings a day, could cost upwards of $400.

      Besides the acute response, it would be helpful to know what the effect on morning ketones is the day after eating meals of various compositions. This is also tricky, since protein, fat and calories may each have their own effect, which would be hard to detangle. For example, if a diet of nothing yields higher ketones than a diet of 70g of protein, it could be an effect of calories. On the other hand, if a diet of 70g of protein yields lower ketones than an isocaloric diet of fat, it is possible that consuming fat itself raises ketones. So you have to be a bit careful.

    2. 2. That's a very interesting idea, and I don't know the answer about uptake proportions. I really like to hear contradictory plausible hypotheses, because it underscores the importance of empirical observation. Thank you.

      It's an important point about the blood sugar being still relatively low, even in the protein fed state. This supports the idea that on a very low carb diet, the total range of blood sugar is much more healthy than achieved when including significant carbohydrate, even in the presence of a lot of protein (at least in non-diabetics).

    3. In rats the level of dietary protein intake affects the level of ketones in subsequent fasting http://www.jbc.org/content/162/1/155.full.pdf - higher protein intake suppresses ketones.

    4. Hi PhilT. Thanks for the paper. That's interesting. It's odd to me that the rats fed the moderately high protein diets were in less deep ketosis than those on a very high carb diet.

      These discussion paragraphs are interesting:

      "MacKay et al. (3) have related the protein intake preceding the fast to the degree of fasting ketosis on the basis of the antiketogenic action of the “stored” protein. The higher nitrogen excretion on fasting after the higher protein intakes was used as a basis for this conclusion. The better maintenance of the blood sugar as well as the liver glycogen level during fasting in those rats was cited as excellent support for the view. The increase in the nitrogen excreted amounted to approximately 6 mg. per sq. dm. per day during the 2 day fast. The antiketogenic material avail- able from the metabolism of protein containing 6 mg. of nitrogen would be equivalent to about 22 mg. of glucose, a very small amount to have such a marked effect upon the metabolism of the fasting animal.

      We have shown previously (1) that a rapid drop in the liver glycogen occurs during the early fasting period in animals previously on a low pro- tein diet and suggested an increased rate of carbohydrate utilization in these animals. Further support for this suggestion was obtained in a recent study (9) in which animals on 5 and 20 per cent protein diets (Nos. 21 and 24) stored 8 and 20 per cent of the caloric intake respectively after similar caloric intakes per unit of body surface. A significantly greater number of calories were utilized by the animals on the low protein diet. The increase in the degree of ketonemia after the low protein diets might be explained by this increased utilization associated with less gluconeo- genesis during fasting. A more rapid loss of the stores of the more readily available antiketogenic material in these animals would account for an earlier increase in the level of the blood ketone bodies. Additional evidence is being accumulated which supports this suggestion."

      I'm not sure, but it seems to suggest that the low-protein animals were in a state of protein deprivation, and therefore the protein-sparing aspect of ketosis was needed more urgently.

      It also seems to say that the protein had been "stored", but *not* as glycogen, because the nitrogen wasn't lost until the fasting began.

      Anyway, lots to ponder.

  3. Great article, very interesting points. Just fishing for some speculation here – about exercise. If increased muscle protein synthesis can redirect some of those nascent amino acids away from gng, then resistance exercise could [theoretically] allow for sustained ketosis at a higher protein intake. So you could [theoretically!] get the benefits of higher protein intake on exercise-induced muscle protein synthesis while maintaining a respectable level of ketosis. Plausible?

    1. Hi, Bill.

      I guess that's built in to the vague notion of "excess" protein. If you do something that increases your protein requirements, like building muscle, then there is less "excess".

      It's an interesting point, because although amino acids that are used for something else can't be used for glucose, there is still a question about metabolic processes that might depend on the total amount of protein consumed regardless of where it ends up going.

      This is related to an argument that several people have brought up both in comments and in personal communication. That is simply that protein must be used for GNG, because if not, then where else does it go? It's a fairly persuasive argument, although not knowing about a pathway doesn't exclude its existence. I read about biological processes happening we thought were impossible fairly regularly. So it's a good argument, but not water-tight.


Note: Only a member of this blog may post a comment.