Keto-adapted, but no ketones?

One of the cheapest and easiest ways to measure ketones is to use ketone test strips, e.g. Ketostix.
Ketone test strips use a chemical reaction to measure acetoacetate (see below), usually in urine, though the same method can be used for blood.
(Not to be confused with the blood strips used at home for beta-hydroxybutyrate.)
However, acetoacetate test strips are of limited usefulness.
For one thing, urine concentrations are affected by dilution, which means that they are affected by how much you drink.
But the problem is deeper than that.
Acetoacetate is only one of the three ketone bodies (see below).
Initially, when you start a ketogenic diet, acetoacetate will make up about half of the circulating ketones [1],
but when you are keto-adapted, it makes up only about 20% of the ketone bodies in circulation (see below).
Morover, the sensitivity of the strips is a little lower than optimal for our purposes.
They register negative unless the concentration is quite high.
So, it is not uncommon for a keto-adapted person to measure negative for acetoacetate.

Different ketone bodies occur in different amounts

There are three compounds grouped together as ketone bodies: acetoacetate, beta-hydroxybutyrate, and acetone.
In keto-adapted people, acetoacetate levels are relatively low even though beta-hydroxybutyrate is high.
Typically, beta-hydroxybutyrate levels are 4–5 times as high as acetoacetate.
(Acetone makes up only about 2% of total ketone bodies [2].)

Beta-hydroxybutyrate and acetoacetate in blood and cerebrospinal fluid during fasting

The graph above shows that in the ketosis of fasting, the proportion of acetoacetate (the top, white part of the bar) is much smaller than that of beta-hydroxybutyrate (the black part).
In the study here, after 21 days of fasting, the average level of blood acetoacetate was 1.04 mmol/L, while the beta-hydroxybutyrate level was 4.95 mmol/L [3].
In another study of epileptic children on ketogenic diets, after 3 months, the average acetoacetate level was 1.182 mmol/L, while the average beta-hydroxybutyrate level was 4.21 [4].
The level of ketosis in fasting and in epileptic treatment is a little bit higher than for the typical ketogenic dieter who is simply trying to lose weight, enhance athletic performance, or improve their cardiovascular risk profile, for example.
In those cases, beta-hydroxybutyrate levels are typically 1–3 mmol/L.
Since the ratio of acetoacetate to beta-hydroxybutyrate is only about 1:4, acetoacetate levels will be only about 0.25–0.75 mmol/L for keto-adapted people.
The acetoacetate measure does not register as positive until about 0.5-1.0 mmol/L [5], so those values will often register as negative for acetoacetate.

Here are some examples of negative acetoacetate, even while beta-hydroxybutyrate is very high.
There is a dangerous state that diabetics can get into called keto-acidosis, which is crucially different from nutritional ketosis (a safe and healthy state), but is often confused with it, because they both involve activation of ketogenesis.
Ketone levels in keto-acidosis are much higher than in nutritional ketosis, and it is the monitoring of this state that ketone strips are optimised for.
Even though ketone levels in keto-acidosis are higher than in nutritional ketosis, in one report it was found that 57% of diabetics with negative acetoacetate measurements were suffering from keto-acidosis [6].

ketosis false negatives using urine acetoacetate
Most of the cases of high beta-hydroxybutyrate in this study were not also positive for urine acetoacetate.
Flowchart to determine diabetic keto-acidosis.
This flowchart shows that it is clinically accepted that even with very high beta-hydroxybutyrate levels, acetoacetate in urine and blood can be negative.
The reason acetoacetate is bothered with at all is that it is relatively cheap and easy to measure.

What’s the best way to measure ketosis?

Ketone test strips are a cheap and easy way to confirm ketosis when you have very high levels, such as during keto-adaptation.
However, we would expect the false negative rate to be high for keto-adapted people, and for infants, (who are normally in consistent but mild ketosis while exclusively breastfed).
So although it can be a good tool when you are starting a ketogenic diet, it is not necessarily reliable as you progress.
A negative acetoacetate measure does not imply that you are not in ketosis.
If you are troubleshooting, and need more accurate measurements, we strongly recommend a blood ketone meter for beta-hydroxybutyrate.
However, be aware that the strips themselves are very expensive.
A new breath acetone meter is now on the market.
It costs about $100, but it doesn’t require any strips, so you pay only once.
Unfortunately, like the acetoacetate strips, the measure is only semi-quantitative, and appears to have a relatively high minimum threshold for showing positive.
We also don’t know how well acetone correlates to beta-hydroxybutyrate, or to therapeutic results.
Nonetheless, it is a promising technology, and it requires no pinpricks or pants down.
We’d love to hear from you if you’ve given it a try.


[1] Evidence type: authority

“Beta-hydroxybutyrate and acetoacetate are made in the liver in about equal proportions, and both are initially promptly oxidized by muscle. But over a matter of weeks, the muscles stop using these ketones for fuel. Instead, muscle cells take up acetoacetate, reduce it to beta-hydroxybutyrate, and return it back into the circulation. Thus after a few weeks, the predominant form in the circulation is beta-hydroxybutyrate, which also happens to be the ketone preferred by brain cells (as an aside, the strips that test for ketones in the urine detect the presence of acetoacetate, not beta-hydroxybutyrate). The result of this process of keto-adaptation is an elegantly choreographed shuttle of fuel from fat cells to liver to muscle to brain.”

[2] Evidence type: authority

Richard A. McPherson, Matthew R. Pincus
Elsevier Health Sciences, Sep 6, 201

“Whenever a defect in carbohydrate metabolism or absorption or an inadequate amount of carbohydrate is present in the diet, the body compensates by metabolizing increasing amounts of fatty acids. […] In ketonuria, the three ketone bodies present in the urine are acetoacetic acid (20%), acetone (2%), and 3-hydroxybutyrate (about 78%).”

[3] Evidence type: experiment

[4] Evidence type: experiment

Neal EG1, Chaffe H, Schwartz RH, Lawson MS, Edwards N, Fitzsimmons G, Whitney A, Cross JH.
Epilepsia. 2009 May;50(5):1109-17. doi: 10.1111/j.1528-1167.2008.01870.x. Epub 2008 Nov 19.

“One hundred forty-five children with intractable epilepsy were randomized to receive a classical or an MCT diet.”
“Classical diets were started at a 2:1 ratio and gradually increased to a 4:1 ratio as tolerated over 1–2 weeks; in a few children the ratio was kept at 3:1 for longer because of tolerance problems. Protein was generally kept at World Health Organization (WHO) minimum requirements for age (World Health Organization, 1985). MCT diets were commenced on a full prescription for carbohydrate (generally 15% energy), protein (usually 10% energy), and long-chain fatty acids (usually 30% energy). The MCT fat was increased incrementally over a 7–10 day period as tolerated, to an initial level that was usually 40–45% of total dietary energy. Diets were fully supplemented with vitamins and minerals.
“Subsequent to starting the diet, all children were reviewed as outpatients at 3, 6, and 12 months. They were also closely monitored by telephone between clinic visits. Diets were fine-tuned as necessary to improve ketosis and optimize seizure control. The parameters within which the two diets could be modified were defined before study commencement. Overall energy prescription was adjusted on both diets as needed. Ketogenic ratio on the classical diets was kept between 2:1 and 5:1 (most classical diet children were on a 4:1 ratio, a few were on a 3:1 ratio, and two children needed a 2:1 ratio for a short period). Fine-tuning on the MCT diets involved adjusting the proportion of MCT and carbohydrate in the prescription. MCT was usually started at 40–45% of energy, and was increased up to 60% if necessary and tolerated. Carbohydrate was usually started at 15% of energy, and was reduced to a lowest value of 12% if necessary. Carbohydrate was reduced to improve ketosis only if an increase in MCT was not possible because of poor tolerance. Other modifications on both diets were fluid intake and meal distribution. Protein intake was increased as needed to meet requirements.”
[5] Evidence type: authority
Ketostrips use nitroprusside to detect acetoacetate levels.
We have seen claims that they can detect as little as 5 mg/dl (0.5 mmol/L), only 10 mg/dl, or, most commonly, the minimum is given as the range 5–10 mg/dl.
Here is an example of each:

Walker HK, Hall WD, Hurst JW, editors. Boston: Butterworths; 1990.

“Nitroprusside is available as a test tablet (Acetest) and as a coated reagent strip (Ketostix), both manufactured by the Ames Division of Miles Laboratories, Inc., Elkhart, Indiana. With Acetest, after 30 seconds the color development is compared to a chart and judged negative, small, moderate or large. The tablet will detect 5 to 10 mg/dl of acetoacetate and 20 mg/dl of acetone. The quantitative range included in each category is 5 to 20 mg/dl for small, 20 to 40 mg/dl for moderate, and 40 mg/dl or greater for large. With Ketostix, the strip is momentarily dipped into the urine specimen or passed through in the urinary stream and compared to a color chart 1 minute later. The scale is negative, trace, small, moderate, and large. The strip is capable of detecting 5 mg/dl acetoacetate but is not reactive to acetone. The ranges are wider and shifted somewhat to the right in the higher zones compared to Acetest so that only 16% of samples containing 20 mg/dl acetoacetate are read as moderate while 24% of samples containing 80 mg/dl acetoacetate are still called moderate. Only 15% of the samples containing 40 mg/dl acetoacetate are judged to be large; 76% are large at 80 mg/dl and 100% at 160 mg/dl. The Ketostix test is most accurate when urines are tested with a high specific gravity (between 1.010 and 1.020) and low-pH. Highly pigmented urine specimens may yield false positive readings. Levodopa will also cause a false positive result. Ketostix strips are less sensitive than Acetest tablets and have a high degree of variability between lots. Acetest, with sensitivity in the 5 mg/dl range, is the preferable method.”

Ochei Et Al. Tata McGraw-Hill Education, Aug 1, 2000. p 134

“Ketostix (Ames)
This test strip will detect 0.5–1.0 mmol/L (5–10 mg/dl) of acetoacetic acid”

Shelly L. Vaden, Joyce S. Knoll, Francis W. K. Smith, Jr., Larry P. Tilley
John Wiley & Sons, Jun 13, 2011

“Only acetoacetate and acetone are detectable by reagent strips or tablet tests, which are based on the reaction of acetoacetate (more reactive) and acetone (less reactive) with nitroprusside.
“Urine (and blood) can be screened for ketones by using either reagent strips or tablets […] The [tablet] is more sensitive than reagent strips and will detect 5 mg/dL of ketones compared with 10 mg/dL for dipsticks.”

[6] Evidence type: authority, since we can’t access the full text

Yutaka Harano, M.D., Masaaki Suzuki, M.D., Hideto Kojima, M.D., Atsunori Kashiwagi, M.D. Ph.D., Hideki Hidaka, M.D. Ph.D. and Yukio Shigeta, M.D. Ph.D.
Diabetes Care September/October 1984 vol. 7 no. 5 481-485

“MacGillivray et al. recently reported that 57% of the urine tests that were negative for ketone bodies by acetest were associated with elevated plasma 3-OHBA in insulin-dependent diabetes.
“MacGillivray, M. H., Voorhess, M. L., Putnam, T. I., Li, P. K., Schaefer, P. A., and Bruck, E.: Hormone and metabolic profiles in children and adolescents with Type I diabetes mellitus. Diabetes Care 1982; 5(Suppl .l):38-47”

The Ketogenic Diet’s Effect on Cortisol Metabolism

(Related post: Red Light, Green Light: responses to cortisol levels in keto vs. longevity research)

One of the myths surrounding ketogenic diets comes from misunderstanding the role of cortisol — the “stress hormone”.

In a previous post, we addressed one of the arguments behind this myth: the idea that to activate gluconeogenesis (to make glucose out of protein), extra cortisol must be recruited.
That is just factually incorrect, as we showed in the post.
The other argument, which we address here, is more complex.
Like the previous cortisol myth, it involves a faulty chain of reasoning.
Here are the steps:

  1. Ketogenic diets may raise certain measures of cortisol.
  2. Chronically elevated cortisol is correlated with metabolic sydrome, and therefore higher cortisol measures may indicate the onset of metabolic syndrome.
  3. Therefore, ketogenic diets could cause metabolic syndrome.

Metabolic syndrome is a terrible and prevalent problem today.
It is that cluster of symptoms most strongly identified with diabetes — excess abdominal fat, high blood sugar, and a particular cholesterol profile — but also correlated with other life-threatening conditions such as heart disease and cancer.
In this post, we’re going to explain some of the specifics of cortisol metabolism.
We’ll show how this argument is vague, and how clarifying it leads to the opposite conclusion.
The confusion may all stem from misunderstanding one important fact:
different measures of cortisol are not equivalent.

First, though, there is an important reason why the argument doesn’t make sense.

We already know that a ketogenic diet effectively treats metabolic syndrome.
As we will describe below, it turns out that certain cortisol patterns are strongly linked to metabolic syndrome, and might even be a cause of metabolic syndrome.
If the cortisol pattern that develops in response to a ketogenic diet were the kind that was associated with metabolic syndrome, then we would expect people on ketogenic diets to show signs of abdominal fat gain, rising blood sugar, and a worsening cholesterol profile, but we see the opposite.
This by itself makes it highly unlikely that ketogenic diets raise cortisol in a harmful way.
In other words, because cortisol regulation is so deeply connected to metabolic syndrome, the fact that ketogenic diets reverse symptoms of metabolic syndrome is itself strong evidence that they improve cortisol metabolism.

In Brief

  • There are many different measures of cortisol, because researchers have identified many different processes in cortisol metabolism.
  • Increases in some of those measurements are consistently linked to metabolic syndrome, and others are not.
  • Some researchers believe that cortisol dysregulation is a key underlying factor in metabolic syndrome.
  • The cornerstone of this connection may be the activity of an enzyme, 11β-HSD1.
    It converts from the inactive form cortisone to the active cortisol.
  • In metabolic syndrome, 11β-HSD1 is underactive in liver tissue and overactive in fat tissue.
    This results in a high rate of cortisol clearance, and low rate of regeneration.
  • These symptoms of cortisol dysregulation associated with metabolic syndrome were found to be reversed by a keto diet in a study that made the necessary measurements.

Does a ketogenic diet raise cortisol?

Boston Children's Hospital graphic (with our markup in black). Click for the original.
Boston Children’s Hospital graphic (with our markup in black). Click for the original.

In a widely-cited study [1], from the Harvard-affiliated Boston Children’s Hospital, published in the Journal of the American Medical Association,
three different diets were tested: a low-fat diet, a low-carb diet, and a low-glycemic-index diet.
The study showed that the different diets had substantially different metabolic effects, with the low-carbohydrate diet having the best results.
To our surprise, the researchers then recommended the low-glycemic-index diet instead.
As they explained in the accompanying press release:

“The very low-carbohydrate diet produced the greatest improvements in metabolism, but with an important caveat: This diet increased participants’ cortisol levels, which can lead to insulin resistance and cardiovascular disease.”

The Boston Children’s Hospital then went on to produce a graphic advising patients to follow the low-glycemic-index diet,
and giving this as the primary reason not to choose the low-carb diet.
Here is that graphic, which we’ve marked (in black) to show our disagreement. (Click for the full version without our markup.)
The cortisol levels are an understandable concern, because high urinary cortisol has been epidemiologically associated with a greatly increased risk of death from heart attacks [2].
However, because a ketogenic diet effectively treats metabolic syndrome, we should expect that it also reduces those specific cortisol patterns that are associated with metabolic syndrome (and therefore heart disease).
As we show below, this has, in fact, been found.

How is cortisol associated with metabolic syndrome?

Figure 1 from “11β-hydroxysteroid dehydrogenase 1: translational and therapeutic aspects.” Gathercole LL, Lavery GG, Morgan SA, Cooper MS, Sinclair AJ, Tomlinson JW, Stewart PM. Endocr Rev. 2013 Aug;34(4):525-55. doi: 10.1210/er.2012-1050. Epub 2013 Apr 23.
Just as we now understand that measuring an individual’s total cholesterol without looking at its component parts is inadequate for assessing cardiovascular health, there are different ways to measure cortisol, and only specific patterns of measurements are found with metabolic syndrome.
Cortisol can be measured in fluids, such as urine, saliva, or blood.
Within those fluids, the amount of free cortisol can be measured, but so can cortisone, the inactive form, or the metabolites that are the result of enzyme action, and the ratios of any of these to the others can be measured
(see Figure 1).
Moreover, these measurements have a diurnal rhythm, being higher and lower at different times of the day.
The enzyme 11β-hydroxysteroid dehydrogenase (11β-HSD) can convert back and forth between cortisol and cortisone.
11β-HSD1—a subtype of 11β-HSD—converts cortisone to cortisol.
When inactive cortisone is converted to the active cortisol, it is called regeneration.
The other enzymes in the illustration break cortisone or cortisol down into metabolites.
That process is called clearance.
It turns out that measurements of these enzyme are important for evaluating cortisol metabolism.
The cortisol profile that has been associated with metabolic syndrome includes the following characteristics:

  • high cortisol production rates [3].
  • high cortisol clearance rates [4], [5].
  • high 11β-HSD1 expression in fat cells, and low 11β-HSD1 expression in the liver [6], [7], which determines when and where cortisol is regenerated.

Similarly to the way total cholesterol measurement is correlated with heart disease, but only because it is roughly correlated with more informative cholesterol measurements, 24-hour urinary cortisol may be a proxy for production or clearance, but a poor one [3], [4], [7].
Cortisol levels are affected by production, but they are also affected by regeneration and clearance.
In other words, if regeneration were increased, or clearance decreased, levels could go up even if production stayed the same or went down.
(We previously discussed a similar situation with blood glucose and faulty inference about glucose production rates.)
This means that levels can look similar, even when cortisol metabolism is very different.

Implication for those following the “adrenal fatigue” hypothesis: if you measure your cortisol, and it is high, you can’t conclude that your adrenal glands are working correspondingly hard. It could be due to increased regeneration and reduced clearance by enzyme activity. Higher cortisol could actually mean the adrenals are working less!
In obesity, it appears that production goes up to compensate for high clearance and impaired regeneration, although sometimes not enough to compensate; blood cortisol is sometimes actually lower in obese subjects [8].

How does a ketogenic diet affect the relevant cortisol measures?

In [9], investigators put obese men on either a high-fat/low-carb (fat 66%, carb 4%) or a moderate-fat/moderate-carb (fat 35%, carb 35%) diet ad libitum (eating as much as they wanted).
Note that both diets had the same protein percent, and both were lower carb than a standard American diet, but only the high-fat/low-carb diet was at ketogenically low levels.
For the high-fat/low-carb group, “the metabolic syndrome pattern” was reversed: blood cortisol went up, clearance went down, and regeneration went up.
This was apparently due to an increase of 11β-HSD1 activity in liver tissue.
(Activity of 11β-HSD1 did not go down in fat tissue of those subjects, but the authors point out that the activity in fat tissue tends to go down when more fat is eaten, and the high-fat/low-carb group weren’t actually eating more fat in absolute terms than at baseline, only lower carb.)
This reversal didn’t happen in the moderate-fat/moderate-carb group, even though they lost a similar amount of weight.
So the ketogenic diet actually improved the cortisol profile of the participants, making it less like the cortisol profile seen in metabolic syndrome.


There is some reason to believe that cortisol dysregulation is a key underlying factor in metabolic syndrome [10], [11].
The dysregulation has a particular pattern that seems to be caused by a tissue-specific expression of the enzyme 11β-HSD1.
There is a belief among some researchers that ketogenic diets worsen cortisol metabolism (which could lead to metabolic syndrome and heart disease),
but an examination of the specific pattern of cortisol metabolism related to metabolic sydrome shows the opposite.
This is what should have been expected in the first place, since ketogenic diets have already been shown to improve insulin sensitivity (the defining symptom of metabolic syndrome) in repeated randomized controlled trials.
One mechanism by which keto diet improves metabolic syndrome may be its beneficial effect on cortisol metabolism.

Further Reading

For a review of 11β-HSD1, see:

Gathercole LL, Lavery GG, Morgan SA, Cooper MS, Sinclair AJ, Tomlinson JW, Stewart PM.
Endocr Rev. 2013 Aug;34(4):525-55. doi: 10.1210/er.2012-1050. Epub 2013 Apr 23.


[1] Evidence type: controlled experiment

Ebbeling CB, Swain JF, Feldman HA, Wong WW, Hachey DL, Garcia-Lago E, Ludwig DS.
JAMA. 2012 Jun 27;307(24):2627-34. doi: 10.1001/jama.2012.6607.

(Emphases ours)
Overweight and obese young adults (n=21).
After achieving 10 to 15% weight loss on a run-in diet, participants consumed low-fat (LF; 60% of energy from carbohydrate, 20% fat, 20% protein; high glycemic load), low-glycemic index (LGI; 40%-40%-20%; moderate glycemic load), and very-low-carbohydrate (VLC; 10%-60%-30%; low glycemic load) diets in random order, each for 4 weeks.

“Hormones and Components of the Metabolic Syndrome (Table 3)
“Serum leptin was highest with the LF diet (14.9 [12.1 to 18.4] ng/mL), intermediate with the LGI diet (12.7 [10.3 to 15.6] ng/mL) and lowest with the VLC diet (11.2 [9.1 to 13.8] ng/mL; P=0.0006). Cortisol excretion measured with a 24-hour urine collection (LF: 50 [41 to 60] μg/d; LGI: 60 [49 to 73] μg/d; VLC: 71 [58 to 86] μg/d; P=0.005) and serum TSH (LF: 1.27 [1.01 to 1.60] μIU/mL; LGI: 1.22 [0.97 to 1.54] μIU/mL; VLC: 1.11 [0.88 to 1.40] μIU/mL; P=0.04) also differed in a linear fashion by glycemic load. Serum T3 was lower with the VLC diet compared to the other two diets (LF: 121 [108 to 135] ng/dL; LGI: 123 [110 to 137] ng/dL; VLC: 108 [96 to 120] ng/dL; P=0.006).
“Regarding components of the metabolic syndrome, indexes of peripheral (P=0.02) and hepatic (P=0.03) insulin sensitivity were lowest with the LF diet. Serum HDL-cholesterol (LF: 40 [35 to 45] mg/dL; LGI: 45 [41 to 50] mg/dL; VLC: 48 [44 to 53] mg/dL; P<0.0001), triglycerides (LF: 107 [87 to 131] mg/dL; LGI: 87 [71 to 106] mg/dL; VLC: 66 [54 to 81] mg/dL; P<0.0001), and PAI-1 (LF: 1.39 [0.94 to 2.05] ng/mL; LGI: 1.15 [0.78 to 1.71] ng/mL; VLC: 1.01 [0.68 to 1.49] ng/mL; P for trend=0.04) were most favorable with the VLC diet and least favorable with the LF diet.

“Although the very low-carbohydrate diet produced the greatest improvements in most metabolic syndrome components examined here, we identified two potentially deleterious effects of this diet. Twenty-four hour urinary cortisol excretion, a hormonal measure of stress, was highest with the very low-carbohydrate diet. Consistent with this finding, Stimson et al reported increased whole-body regeneration of cortisol by 11β-HSD1 and reduced inactivation of cortisol by 5α-and 5β-reductases over 4 weeks on a VLC vs. a moderate-carbohydrate diet. Higher cortisol levels may promote adiposity, insulin resistance, and cardiovascular disease, as observed in epidemiological studies.”

Comment: It is ironic that the authors bring up Stimson et al. as an example of a study that corroborates their findings. This is the very study [9] that, in our opinion, exonerates the VLC diet with respect to cortisol.

[2] Evidence type: epidemiological observation

Vogelzangs N, Beekman AT, Milaneschi Y, Bandinelli S, Ferrucci L, Penninx BW.
J Clin Endocrinol Metab. 2010 Nov;95(11):4959-64. doi: 10.1210/jc.2010-0192. Epub 2010 Aug 25.

“Context: The stress hormone cortisol has been linked with unfavorable cardiovascular risk factors, but longitudinal studies examining whether high levels of cortisol predict cardiovascular mortality are largely absent.
Objective: The aim of this study was to examine whether urinary cortisol levels predict all-cause and cardiovascular mortality over 6 yr of follow-up in a general population of older persons.
Design and Setting: Participants were part of the InCHIANTI study, a prospective cohort study in the older general population with 6 yr of follow-up.
Participants: We studied 861 participants aged 65 yr and older.
Main Outcome Measure: Twenty-four-hour urinary cortisol levels were assessed at baseline. In the following 6 yr, all-cause and cardiovascular mortality was ascertained from death certificates. Cardiovascular mortality included deaths due to ischemic heart disease and cerebrovascular disease.
Results: During a mean follow-up of 5.7 (sd = 1.2) yr, 183 persons died, of whom 41 died from cardiovascular disease. After adjustment for sociodemographics, health indicators, and baseline cardiovascular disease, urinary cortisol did not increase the risk of noncardiovascular mortality, but it did increase cardiovascular mortality risk. Persons in the highest tertile of urinary cortisol had a five times increased risk of dying of cardiovascular disease (hazard ratio = 5.00; 95% confidence interval = 2.02–12.37). This effect was found to be consistent across persons with and without cardiovascular disease at baseline (p interaction = 0.78).
Conclusions: High cortisol levels strongly predict cardiovascular death among persons both with and without preexisting cardiovascular disease. The specific link with cardiovascular mortality, and not other causes of mortality, suggests that high cortisol levels might be particularly damaging to the cardiovascular system.”
[3] Evidence type: experiment

Jonathan Q. Purnell, Steven E. Kahn, Mary H. Samuels, David Brandon, D. Lynn Loriaux, and John D. Brunzell
Am J Physiol Endocrinol Metab. 2009 February; 296(2): E351–E357.

“Controversy exists as to whether endogenous cortisol production is associated with visceral obesity and insulin resistance in humans. We therefore quantified cortisol production and clearance rates, abdominal fat depots, insulin sensitivity, and adipocyte gene expression in a cohort of 24 men. To test whether the relationships found are a consequence rather than a cause of obesity, eight men from this larger group were studied before and after weight loss. Daily cortisol production rates (CPR), free cortisol levels (FC), and metabolic clearance rates (MCR) were measured by stable isotope methodology and 24-h sampling; intra-abdominal fat (IAF) and subcutaneous fat (SQF) by computed tomography; insulin sensitivity (SI) by frequently sampled intravenous glucose tolerance test; and adipocyte 11β-hydroxysteroid dehydrogenase-1 (11β-HSD-1) gene expression by quantitative RT-PCR from subcutaneous biopsies. Increased CPR and FC correlated with increased IAF, but not SQF, and with decreased SI. Increased 11β-HSD-1 gene expression correlated with both IAF and SQF and with decreased SI. With weight loss, CPR, FC, and MCR did not change compared with baseline; however, with greater loss in body fat than lean mass during weight loss, both CPR and FC increased proportionally to final fat mass and IAF and 11β-HSD-1 decreased compared with baseline. These data support a model in which increased hypothalamic-pituitary-adrenal activity in men promotes selective visceral fat accumulation and insulin resistance and may promote weight regain after diet-induced weight loss, whereas 11β-HSD-1 gene expression in SQF is a consequence rather than cause of adiposity.
“Previous studies have shown that compared with women, men have increased CPR (29), cortisol levels (29, 44), and visceral adiposity (9, 13). Given that hypercortisolemia can induce central obesity in disease states such as Cushing’s syndrome, elevated endogenous cortisol secretion has been considered a potential mechanism that contributes to the expression of visceral adiposity in humans. However, of four previous reports that used 24-h urinary excretion rates of cortisol as a surrogate for cortisol production, only one found significant relationships between urinary secretion of total glucocorticoids, truncal fat, and insulin sensitivity in men and women (39), while three other studies in men have failed to show associations between urinary glucocorticoid secretion and either WHR (16, 26) or visceral fat (48). These studies, however, did not measure cortisol production directly, did not include blood FC, and did not test for differences in circadian variations of blood levels of cortisol, and in only one study was visceral fat specifically measured.
“In summary, we found in men that increased CPR and circulating FC are associated with accumulation of IAF, but not SQF, and with insulin resistance and impaired islet β-cell compensation (DI).”

[4] Evidence type: observational

Holt HB, Wild SH, Postle AD, Zhang J, Koster G, Umpleby M, Shojaee-Moradie F, Dewbury K, Wood PJ, Phillips DI, Byrne CD.
Diabetologia. 2007 May;50(5):1024-32. Epub 2007 Mar 17.

The regulation of cortisol metabolism in vivo is not well understood. We evaluated the relationship between cortisol metabolism and insulin sensitivity, adjusting for total and regional fat content and for non-alcoholic fatty liver disease.
“Twenty-nine middle-aged healthy men with a wide range of BMI were recruited. We measured fat content by dual-energy X-ray absorptiometry and magnetic resonance imaging (MRI), liver fat by ultrasound and MRI, the hypothalamic-pituitary-adrenal axis by adrenal response to ACTH(1-24), unconjugated urinary cortisol excretion, corticosteroid-binding globulin, and cortisol clearance by MS. We assessed insulin sensitivity by hyperinsulinaemic-euglycaemic clamp and by OGTT.
“Cortisol clearance was strongly inversely correlated with insulin sensitivity (M value) (r = -0.61, p = 0.002). Cortisol clearance was increased in people with fatty liver compared with those without (mean+/-SD: 243 +/- 10 vs 158 +/- 36 ml/min; p = 0.014). Multiple regression modelling showed that the relationship between cortisol clearance and insulin sensitivity was independent of body fat. The relationship between fatty liver and insulin sensitivity was significantly influenced by body fat and cortisol clearance.
“Cortisol clearance is strongly associated with insulin sensitivity, independently of the amount of body fat. The relationship between fatty liver and insulin sensitivity is mediated in part by both fatness and cortisol clearance.”
“Since we showed no strong associations between measures of insulin sensitivity and 09.00 h cortisol levels, ACTH-stimulated cortisol concentrations, and unconjugated urinary cortisol excretion, these findings suggest that the relationship between these other aspects of cortisol metabolism and insulin sensitivity is relatively weak.”

[5] Evidence type: experiment

(emphasis ours)
“The present study was designed to examine the hypothesis that hypothalamic-pituitary-adrenal axis activity as measured by 24-h cortisol production rate (CPR) and plasma levels of free cortisol is linked to increased body fat in adults, and that increased cortisol levels with aging results from increased CPR. Fifty-four healthy men and women volunteers with a wide range of body mass indexes and ages underwent measurement of CPR by isotope dilution measured by gas chromatography-mass spectroscopy, cortisol-binding globulin, and free cortisol in pooled 24-h plasma, body composition, and leptin. Cortisol clearance rates were determined from the 10-h disappearance curves of hydrocortisone after steady-state infusion in a separate group of lean and obese subjects with adrenal insufficiency. Although CPR significantly increased with increasing body mass index and percentage body fat, free cortisol levels remained independent of body composition and leptin levels due to increased cortisol clearance rates. CPR and free cortisol levels were, however, significantly higher in men than women. In addition, 24-h plasma free cortisol levels were increased with age in association with increased CPR, independent of body size. This increase in hypothalamic-pituitary-adrenal axis activity may play a role in the alterations in body composition and central fat distribution in men vs. women and with aging.”

[6] Evidence type: review of human and non-human animal experiments

Espíndola-Antunes D, Kater CE.
Arq Bras Endocrinol Metabol. 2007 Nov;51(8):1397-403.

“Human studies
“The bulk of evidences points both to an overexpression and an increased activity of 11bHSD1 in subcutaneous (SAT) and visceral adipose tissue (VAT) of obese subjects, although biopsies of the omentum were conducted in but a few studies. Several groups have shown higher 11bHSD1 mRNA expression in obese compared to non-obese subjects (29-32), although not all studies agree (33). Direct in vivo measurements using microdialysis in SAT also suggest an increase in the conversion rate of cortisone to cortisol (34). Moreover, 11bHSD1 mRNA expression positively correlates with obesity (body mass index and abdominal circumference), body composition, insulin resistance (30-32), resistins and other cytokines, as TNFa, IL-6, and leptin (35).
“The whole body 11bHSD1 activity reflects mainly hepatic expression. Initial studies that relied on measurements of cortisol-to-cortisone metabolites in urine (23,36) should be taken with caution as indicative of 11bHSD1 activity, because several other cortisol and cortisone metabolizing enzymes are deregulated in obesity (36). Of greater importance is the finding of reduced hepatic 11bHSD1 activity measured by the conversion of orally administered cortisone to cortisol (23,37). Thus, 11bHSD1 upregulation in obesity seems not to be a generalized process. In both the whole body and the splanchnic circulation there are no differences between obese and lean subjects regarding cortisol regeneration rates (as measured by [2H4]-cortisol tracer), presumably because an upregulation in adipose tissue is counterbalanced by a downregulation in the liver (15).
“Polymorphisms in the 11bHSD1 gene were identified in an attempt to clarify the basis for the increased activity of adipose tissue 11bHSD1 in obesity. In two populations, polymorphisms were associated with an increased risk of diabetes and hypertension, but not obesity (38,39). A polymorphism was also found that predicts lower 11bHSD1 expression and protection against diabetes (40).”

[7] Evidence type: observational

Wake DJ, Rask E, Livingstone DE, Söderberg S, Olsson T, Walker BR.
J Clin Endocrinol Metab. 2003 Aug;88(8):3983-8.

(emphasis ours)
“In idiopathic obesity circulating cortisol levels are not elevated, but high intraadipose cortisol concentrations have been implicated. 11beta-Hydroxysteroid dehydrogenase type 1 (11HSD1) catalyzes the conversion of inactive cortisone to active cortisol, thus amplifying glucocorticoid receptor (GR) activation. In cohorts of men and women, we have shown increased ex vivo 11HSD1 activity in sc adipose tissue associated with in vivo obesity and insulin resistance. Using these biopsies, we have now validated this observation by measuring 11HSD1 and GR mRNA and examined the impact on intraadipose cortisol concentrations, putative glucocorticoid regulated adipose target gene expression (angiotensinogen and leptin), and systemic measurements of cortisol metabolism. From aliquots of sc adipose biopsies from 16 men and 16 women we extracted RNA for real-time PCR and steroids for immunoassays. Adipose 11HSD1 mRNA was closely related to 11HSD1 activity [standardized beta coefficient (SBC) = 0.58; P < 0.01], and both were positively correlated with parameters of obesity (e.g. for BMI, SBC = 0.48; P < 0.05 for activity, and SBC = 0.63; P < 0.01 for mRNA) and insulin sensitivity (log fasting plasma insulin; SBC = 0.44; P < 0.05 for activity, and SBC = 0.33; P = 0.09 for mRNA), but neither correlated with urinary cortisol/cortisone metabolite ratios. Adipose GR-alpha and angiotensinogen mRNA levels were not associated with obesity or insulin resistance, but leptin mRNA was positively related to 11HSD1 activity (SBC = 0.59; P < 0.05) and tended to be associated with parameters of obesity (BMI: SBC = 0.40; P = 0.09), fasting insulin (SBC = 0.65; P < 0.05), and 11HSD1 mRNA (SBC = 0.40; P = 0.15). Intraadipose cortisol (142 +/- 30 nmol/kg) was not related to 11HSD1 activity or expression, but was positively correlated with plasma cortisol. These data confirm that idiopathic obesity is associated with transcriptional up-regulation of 11HSD1 in adipose, which is not detected by conventional in vivo measurements of urinary cortisol metabolites and is not accompanied by dysregulation of GR. Although this may drive a compensatory increase in leptin synthesis, whether it has an adverse effect on intraadipose cortisol concentrations and GR-dependent gene regulation remains to be established.”

[8] Evidence type: review

(emphasis ours)
“The parallels between the clinical features of Cushing’s syndrome and the features of the metabolic syndrome (visceral obesity, hyperglycaemia, hypertension) led to the hypothesis that obesity is associated with glucocorticoid excess (Bjorntorp, 1991). In several monogenic rodent models obesity is accompanied by elevated circulating glucocorticoid levels, and the obesity is prevented by adrenalectomy. Hyperactivity of the HPA axis was thought to reflect chronic stress. However, although there is some evidence for greater stress responsiveness of the HPA axis in obesity (Rosmond et al . 1998; Epel et al . 2000), stress does not appear to explain HPA axis activation in metabolic syndrome (Brunner et al . 2002). Most importantly, in human obesity it appears that cortisol secretion (Marin et al . 1992; Pasquali et al . 1993) is increased to match elevated metabolic clearance (Strain et al . 1982; Andrew et al . 1998; Lottenberg et al . 1998), and does not result in increased plasma cortisol concentrations. Indeed, plasma cortisol concentrations are generally lower amongst obese subjects (Ljung et al . 1996; Walker et al . 2000; Reynolds et al . 2003), i.e. inverse to the effects on the HPA axis seen during starvation (see earlier; p. 2).”

[9] Evidence type: randomised controlled trial

Stimson RH, Johnstone AM, Homer NZ, Wake DJ, Morton NM, Andrew R, Lobley GE, Walker BR.
J Clin Endocrinol Metab. 2007 Nov;92(11):4480-4. Epub 2007 Sep 4.

Dietary macronutrient composition influences cardiometabolic health independently of obesity. Both dietary fat and insulin alter glucocorticoid metabolism in rodents and, acutely, in humans. However, whether longer-term differences in dietary macronutrients affect cortisol metabolism in humans and contribute to the tissue-specific dysregulation of cortisol metabolism in obesity is unknown.
The objective of the study was to test the effects of dietary macronutrients on cortisol metabolism in obese men.
DESIGN: The study consisted of two randomized, crossover studies.
SETTING: The study was conducted at a human nutrition unit.
PARTICIPANTS: Participants included healthy obese men.
INTERVENTIONS, OUTCOME MEASURES, AND RESULTS: Seventeen obese men received 4 wk ad libitum high fat-low carbohydrate (HF-LC) (66% fat, 4% carbohydrate) vs. moderate fat-moderate carbohydrate (MF-MC) diets (35% fat, 35% carbohydrate). Six obese men participated in a similar study with isocaloric feeding. Both HF-LC and MF-MC diets induced weight loss. During 9,11,12,12-[(2)H](4)-cortisol infusion, HF-LC but not MF-MC increased 11beta-hydroxysteroid dehydrogenase type 1 (11beta-HSD1) activity (rates of appearance of cortisol and 9,12,12-[(2)H](3)-cortisol) and reduced urinary excretion of 5alpha- and 5beta-reduced [(2)H](4)-cortisol metabolites and [(2)H](4)-cortisol clearance. HF-LC also reduced 24-h urinary 5alpha- and 5beta-reduced endogenous cortisol metabolites but did not alter plasma cortisol or diurnal salivary cortisol rhythm. In sc abdominal adipose tissue, 11beta-HSD1 mRNA and activity were unaffected by diet.
CONCLUSIONS: A low-carbohydrate diet alters cortisol metabolism independently of weight loss. In obese men, this enhances cortisol regeneration by 11beta-HSD1 and reduces cortisol inactivation by A-ring reductases in liver without affecting sc adipose 11beta-HSD1. Alterations in cortisol metabolism may be a consequence of macronutrient dietary content and may mediate effects of diet on metabolic health.”

[10] Evidence type: authority (review article)

Pereira CD, Azevedo I, Monteiro R, Martins MJ.
Diabetes Obes Metab. 2012 Oct;14(10):869-81. doi: 10.1111/j.1463-1326.2012.01582.x. Epub 2012 Mar 8.

(emphasis ours)
Recent evidence strongly argues for a pathogenic role of glucocorticoids and 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1) in obesity and the metabolic syndrome, a cluster of risk factors for atherosclerotic cardiovascular disease and type 2 diabetes mellitus (T2DM) that includes insulin resistance (IR), dyslipidaemia, hypertension and visceral obesity. This has been partially prompted not only by the striking clinical resemblances between the metabolic syndrome and Cushing’s syndrome (a state characterized by hypercortisolism that associates with metabolic syndrome components) but also from monogenic rodent models for the metabolic syndrome (e.g. the leptin-deficient ob/ob mouse or the leptin-resistant Zucker rat) that display overall increased secretion of glucocorticoids. However, systemic circulating glucocorticoids are not elevated in obese patients and/or patients with metabolic syndrome. The study of the role of 11β-HSD system shed light on this conundrum, showing that local glucocorticoids are finely regulated in a tissue-specific manner at the pre-receptor level. The system comprises two microsomal enzymes that either activate cortisone to cortisol (11β-HSD1) or inactivate cortisol to cortisone (11β-HSD2). Transgenic rodent models, knockout (KO) for HSD11B1 or with HSD11B1 or HSD11B2 overexpression, specifically targeted to the liver or adipose tissue, have been developed and helped unravel the currently undisputable role of the enzymes in metabolic syndrome pathophysiology, in each of its isolated components and in their prevention. In the transgenic HSD11B1 overexpressing models, different features of the metabolic syndrome and obesity are replicated.”

[11] Evidence type: non-human animal experiment

Schnackenberg CG, Costell MH, Krosky DJ, Cui J, Wu CW, Hong VS, Harpel MR, Willette RN, Yue TL.
Biomed Res Int. 2013;2013:427640. doi: 10.1155/2013/427640. Epub 2013 Mar 18.

(emphasis ours)
“Metabolic syndrome is a constellation of risk factors including hypertension, dyslipidemia, insulin resistance, and obesity that promote the development of cardiovascular disease. Metabolic syndrome has been associated with changes in the secretion or metabolism of glucocorticoids, which have important functions in adipose, liver, kidney, and vasculature. Tissue concentrations of the active glucocorticoid cortisol are controlled by the conversion of cortisone to cortisol by 11β-hydroxysteroid dehydrogenase type 1 (11β-HSD1). Because of the various cardiovascular and metabolic activities of glucocorticoids, we tested the hypothesis that 11β-HSD1 is a common mechanism in the hypertension, dyslipidemia, and insulin resistance in metabolic syndrome. In obese and lean SHR/NDmcr-cp (SHR-cp), cardiovascular, metabolic, and renal functions were measured before and during four weeks of administration of vehicle or compound 11 (10 mg/kg/d), a selective inhibitor of 11β-HSD1. Compound 11 significantly decreased 11β-HSD1 activity in adipose tissue and liver of SHR-cp. In obese SHR-cp, compound 11 significantly decreased mean arterial pressure, glucose intolerance, insulin resistance, hypertriglyceridemia, and plasma renin activity with no effect on heart rate, body weight gain, or microalbuminuria. These results suggest that 11β-HSD1 activity in liver and adipose tissue is a common mediator of hypertension, hypertriglyceridemia, glucose intolerance, and insulin resistance in metabolic syndrome.

Science Fiction

We humans are storytellers.
When we want something to be memorable and meaningful, we make it into a story that can be interpreted causally.
Our brains are just made that way [1], [2].
That may be why adding dramatic anecdotes to a book or article can make it more popular and persuasive than sticking solely to claims that have gone through rigorous tests.
This can lead us to believe more strongly in a hypothesis than the evidence merits.
In Apologia we describe how we try to mitigate errors in our work by tracing our sources of evidence.
We aim to at least label our hypotheses as such, and to point out when evidence we used comes only from observed correlations.
Beyond anecdote, there are other ways that storytelling is used in science: the plausible mechanism, and the evolutionary story.
Both types are used in two ways: to explain an observation, or to lend weight to an hypothesis.
A characteristic of this “science fiction” is that it can be argued against by making a different, more compelling story.
The best narrator wins!
Like any other hypothesis, though, an hypothesis supported by plausible mechanisms or evolutionary stories might be disproven by experimental evidence.

Plausible mechanisms

Plausible mechanisms are stories that involve chains of known scientific facts, integrated together into an expected result.

Here’s an example of plausible mechanism that we used in a post a couple of years ago:

  1. “[BCAAs] are known to have positive effects on muscle growth and recovery.
  2. “One important effect of keto-adaptation is a dramatic increase in circulating BCAAs.”
  3. “Therefore it is quite plausible that […] a ketogenic diet will improve muscle growth and recovery relative to a glycolytic diet, something already anecdotally reported.”

(We’re glad to see that we drew attention to the type of argument we were making, with the word plausible, even as we threw in anecdote for good measure.)
A counter-story might go along these lines:

  1. Muscle growth is stimulated by insulin.
  2. Ketogenic diets lower insulin.
  3. Therefore muscle growth will be retarded on a ketogenic diet in comparison with a glycolytic diet.

We could spend a lot of energy expounding on one story or other, but only replicable, randomised, controlled trials can finally lay a story to rest.

Evolutionary stories

Evolutionary stories are stories that make plausible hypotheses about why certain adaptations could have been selected.
They can be used to argue that a certain trait is adaptive, or, more subtly, that an organism has a certain trait.

Here’s an example of an evolutionary story:

  1. Ketogenic metabolism is an adaptation to cope with conditions of scarcity.
  2. We get fat because our genes are thriftily hanging on to excess calories to use in famines.
  3. Therefore ketogenic diets must send stress signals to the body, indicating imminent famine.

We recently read a counter-story described by Anna of the blog “lifeextension”.
We can’t do that whole post justice, so we recommend you read it.
It’s not long, and it contains citations supporting the premises.
The gist of the story goes like this:

  1. We adapted to a diet largely consisting of meat.
  2. When meat is plentiful, our bodies keep fat at optimal levels for peak fitness and ability.
  3. Eating starch or fiber sends a strong stress signal, mediated by gut microbiota which are sensitive to small changes in diet, indicating that famine is imminent.
  4. This starts a cascade of fat storage.
  5. That is, starch digestion is an adaptation to cope with conditions of scarcity.

We could add more to that story.
For example,

  • There is a follow-up post about work that demonstrates that amylase activity is a biomarker of stress.
  • We could point out that no other species responds to abundance by getting fat. Instead they reproduce more.

But these are just embellishments to one of the stories. The right question to ask is:

What kind of evidence could let us discriminate between these two stories?

One way to do that would be to understand stress, how to measure it, and how to tell when it is healthy and when it is detrimental.
Tune in next time for “Ketogenic diets, Cortisol, and Stress, Part II”: The Ketogenic Diet’s Effect on Cortisol Metabolism, in which we provide evidence that ketogenic diets have a beneficial effect on cortisol metabolism.

Further Reading

  1. Darwin’s Dangerous Idea by Daniel Dennett
    In which we are warned against “Just So Stories”—evolutionary stories that are plausible (at least to some people), and that would explain some observations, but that may or may not be true.
  2. Just So Stories by Rudyard Kipling
    Stories explaining how animals came to be the way they are, such as “How The Leopard Got His Spots”.
  3. The Black Swan by Nassim Nicholas Taleb
    In which we are warned against “The Narrative Fallacy“—the tendency to rely on explanations because they make good stories.
  4. Thrifty genes for obesity and the metabolic syndrome — time to call off the search?
  5. Salivary alpha-amylase in biobehavioral research: recent developments and applications.


[1] The Neurology of Narrative

Kay Young, Jeffrey L. Saver
From: SubStance Issue 94/95 (Volume 30, Number 1&2), 2001 pp. 72-84

Narrative is the inescapable frame of human existence. Thinkers as diverse as Aristotle, Barthes, and Bruner have recognized the centrality of narrative in human cognition, but have scanted its neurobiologic underpinning. Recent advances in cognitive neuroscience suggest a regionally distributed neural network mediates the creation of narrative in the human central nervous system. Fundamental network components include: 1) the amygdalo-hippocampal system, responsible for initial encoding of episodic and autobiographical memories, 2) the left peri-Sylvian region, where language is formulated, and 3) the frontal cortices and their subcortical connections, where individuals and entities are organized into real and fictional temporal narrative frames. We describe four types of dysnarrativia, states of narrative impairment experienced by individuals with discrete focal damage in different regions of this neural network subserving human self-narrative. Patients with these syndromes illustrate the inseparable connection between narrativity and personhood. Brain- injured individuals may lose their linguistic or visuospatial competencies and still be recognizably the same persons. Individuals who have lost the ability to construct narrative, however, have lost their selves.”

[2] Causal coherence and memory for events in narratives

John B. Black Hyman Bern
Journal of Verbal Learning and Verbal Behavior. Volume 20, Issue 3, June 1981, Pages 267–275

“Causally related events in narratives were remembered better than events that were not causally related. In Experiment 1, subjects recalled sentences from stories when given the sentences that immediately preceded them as cues. Cued recall was better when the two sentences were causally related than when they were not. In Experiment 2, subjects free recalled the same stories. Again, recall was better for the sentences when they were part of a causally related pair. Also subjects were more likely to combine two sentences into one during recall when they were causally related.”

Babies thrive under a ketogenic metabolism

Some people, even some scientists who study ketogenic metabolism, have the idea that ketogenesis is somehow abnormal, or exceptional; an adaptation for emergencies only.
We disagree.
One reason we think a ketogenic metabolism is normal and desirable, is that human newborns are in ketosis.
Despite the moderate sugar content of human breast milk, breastfeeding is particularly ketogenic.
This period of development is crucial, and there is extensive brain growth during it.
Although the composition of breast milk can be affected by diet [1], it is reasonable to assume that breast milk has always been ketogenic, and this is not an effect of modernisation.
When the brain is in its period of highest growth, and when the source of food is likely to be close to what it evolved to be for that period, ketones are used to fuel that growth.
If nothing else, this suggests that learning is well supported by a ketogenic metabolism.
It is also consistent with the ability of ketogenic diets to treat a variety of seemingly unrelated brain disorders and brain trauma.

In brief

  • Newborn infants are in ketosis.
    This is their normal state.
  • Breastfeeding is particularly ketogenic (compared to formula feeding).
  • Breastfeeding longer (up to a point) is associated with better health outcomes.
  • This suggests the hypothesis that weaning onto a ketogenic diet would be healthier than weaning onto a high-carb diet.

(Mark-up ours)

Human babies are in ketosis

Soon after birth, human babies are in ketosis, and remain so while breastfeeding [2].
They use ketones and fats for energy and for brain growth.
When this has been studied,
in the first couple of hours after birth, babies aren’t immediately in ketosis.
There is a short delay [3].
During that brief period before ketogenesis starts, lactate (confusingly not to do with lactation) becomes an important fuel to suppport the brain [4].
Some researchers speculate that this delay in ketogenesis could be because of a limited supply of carnitine, which is supplied by milk,
but they also note that glycogenolysis and gluconeogenesis (the process by which glucose is made out of protein) are not active immediately [5].
Therefore, it could simply be the case that ketogenesis takes time to get started.
In other words, it may just be keto-adaptation.
Note, though, that the mothers of these babies were unlikely to have been ketogenic.
As it happens, if the mother is in ketosis (as has been studied through fasting), ketone bodies will pass through the placenta and be used by the fetus [5], [6].
At the same time, gluconeogenesis is induced in the liver of the fetus, likely as a result of the insulin-to-glucagon ratio [7], [8].
Therefore, it is possible that the fetus of a ketogenic mother would already be independently ketogenic at birth.

Breastfeeding is probably healthy

Many positive associations between exclusive breastfeeding for at least 3-6 months and the later health of the child have been reported.
For example, intelligence has been positively correlated with length of time breastfeeding.
The data is conflicting and prone to confounds [9], although we found a few studies that appear to have addressed those confounds and still showed an effect [10], [11], [12].
There have also been correlations found between breastfeeding and protection from developing diseases, such as asthma and allergies [13], type 1 and type 2 diabetes [14], and epilepsy [15].
Observational correlations are good sources of hypotheses, but can’t establish causality.
Unfortunately, these hypotheses are hard to test.
We suppose that breastfeeding is healthy mainly because we clearly evolved to breastfeed.

Breastfeeding is ketogenic

The medical focus in the 20th century was heavily influenced by the discovery of micronutrients, and because of this, we have been looking for the secret of the healthfulness of breast milk by examining what nutrients it contains.
However, one significant difference between breastfeeding infants and those drinking formula is that they are in deeper ketosis [16].
It is not known why.
It could be a property of the milk, or something else about the feeding.
In any case, regardless of mechanism, the fact is that breastfeeding is more ketogenic.
It is possible that the reason that longer breastfeeding is generally associated with better health, is because it represents a longer time in ketosis.


  • The period in which human brains grow the most, and in which food is least likely to be different from evolutionary conditions, is a ketogenic period.
    This suggests that a ketogenic metabolism is excellent for learning and development.
  • Breastfeeding in humans is particularly ketogenic.
    We hypothesise that the positive associations between health and longer breastfeeding may be due to extending the period of ketosis in infancy.
  • A related hypothesis we offer is that extending the period of ketosis after breastfeeding, by weaning onto ketogenic foods such as homemade broth [*] and fatty meat, rather than cereal, fruit, and starchy vegetables, would further promote brain development and reduce risk of disease.
[*] Homemade, because it is rich in fat, unlike the boxed varieties which have almost none.

References and notes

[1] Evidence type: review

Sheila M. Innis
Adv Nutr May 2011 Adv Nutr vol. 2: 275-283, 2011

“The fatty acids needed by the mammary gland for synthesis of TG for secretion in milk are obtained by uptake of fatty acids from plasma and de novo synthesis in the mammary gland (7). Fatty acid synthesis in the mammary gland, however, is unusual. Commencing with acetyl CoA, malonyl CoA 2 carbon units are added to the growing fatty acid with elongation terminated at a carbon chain length of 14 or less by the mammary gland-specific enzyme thioesterase II rather than at 16 carbons, as occurs in the liver and other tissues (7, 8). Synthesis and secretion of 10:0, 12:0, and 14:0 into milk is increased in lactating women consuming high-carbohydrate diets, whereas the secretion of the 18 carbon chain unsaturated fatty acids, which are derived by uptake from plasma, is decreased (4, 9, 10). Overall, reciprocal changes in mammary gland-derived medium-chain fatty acids (MCFA) and plasma-derived unsaturated fatty acids allows the milk fat content to be maintained under conditions of varying maternal dietary fat and carbohydrate intake. The levels of unsaturated fatty acids, including 18:1(n-9), 18:2(n-6), 18:3(n-3), 20:5(n-3), 22:6(n-3), and trans fatty acids in human milk, however, vary widely, with the maternal dietary fat composition being one of the most important factors contributing to the differences in the levels of unsaturated fatty acids in the milk of different women (4, 8, 11–14). In contrast, the levels of 16:0 in milk from women in different countries and with different diets is relatively constant at 20–25% of the milk fatty acids regardless of differences in the maternal diet fat content or composition (4). Possible exceptions include lower levels of 14–18% 16:0 described for milk from women in Gambia (15), some vegans and vegetarians (16, 17), and the Arctic Inuit (18).”

[2] Evidence type: review of experiments in humans and rats

Medina JM, Tabernero A.
J Neurosci Res. 2005 Jan 1-15;79(1-2):2-10.

“Striking changes in the fuel supply to the tissues occur during the perinatal period because the transplacental supply of nutrients ends with a period of postnatal starvation (presuckling period) followed by adaptation to a fat-rich diet.”
“Ketone bodies are a major fuel for the brain during the suckling period and hence the stimulation of ketogenesis at birth is an important metabolic event in adaptation of the newborn to extrauterine life. Ketogenesis is active during late gestation in human fetal liver and the activity of ketogenic enzymes sharply increases immediately after birth in the rat (Hahn and Novak, 1985; Bougneres et al., 1986). In addition to modulation of enzyme activities, the control of ketogenesis also depends on the availability of fatty acids. The increase in fatty acid concentrations that occurs after delivery is due to breakdown of triacylglycerol in white adipose tissue present in human newborns at birth. In the rat, however, plasma fatty acids mostly come from hydrolysis of triacylglycerols from the mother’s milk because of the lack of white adipose tissue at birth. Nevertheless, in both species, once lactation is active fatty acids come from the intestinal hydrolysis of milk triacylglycerols, which may be absorbed directly without passage through the lymph (Aw and Grigor, 1980).”
“The increase in the activities of ketogenic enzymes together with the increase in the availability of fatty acids occurring immediately after delivery result in enhancement of ketogenic capacity of the liver (Girard,1990). This is responsible for the increase in ketone body concentrations observed postnatally. In fact, plasma ketone body concentrations are the main factor controlling the rate of ketone body utilization by neonatal tissues (Robinson and Williamson, 1980). In addition, activities of enzymes involved in ketone body utilization either increase during the first days of extrauterine life, as in the rat (Page et al., 1971), or are already induced during early gestation, as in the human brain (Patel et al., 1975). Moreover, newborn rat brain contains acetoacetyl-CoA synthetase, a unique enzyme that allows an important portion of carbon atoms from ketone bodies to be incorporated into lipid via a highly efficient cytosolic pathway (Williamson and Buckley, 1973). Indeed, there is a strong correlation between lipid synthesis and the activity of this enzyme during brain development (Yeh and Sheehan, 1985). Moreover, ketone body transport across the blood–brain barrier using the monocarboxylate carrier is maximal during the suckling period, in keeping with the idea that ketone bodies play an important role in brain development (Cremer, 1982; Conn et al., 1983).
“Ketone bodies are utilized by the newborn brain as a source of energy and carbon skeletons and are incorporated into fatty acids, sterols, acetylcholine, and amino acids (Robinson and Williamson, 1980; Bougneres et al., 1986). Ketone bodies, however, seem to be the major source of carbon skeletons for sterol synthesis during brain development and play a decisive role in the synthesis of brain structures during myelinogenesis (Robinson and Williamson, 1980; Miziorko et al., 1990). Ketone bodies are utilized evenly by neurons, astrocytes, and oligodendrocytes (Edmond et al., 1987; Lopes-Cardozo et al., 1989; Poduslo and Miller, 1991), indicating that they are ubiquitous substrates for brain cells. Acetoacetyl-CoA synthetase activity, however, is higher in oligodendrocytes than in neurons or astrocytes, confirming the special role of oligodendrocytes in myelinogenesis (Pleasure et al., 1979; Lopes-Cardozo et al., 1989; Poduslo and Miller, 1991).”

[3] Evidence type: review of experiments

Ward Platt M, Deshpande S.
Semin Fetal Neonatal Med. 2005 Aug;10(4):341-50.

“During the first 8 h after birth, newborn infants have been shown to have rather low plasma ketone body concentrations despite adequate levels of precursor free fatty acids (FFAs), reflecting limited capacity for hepatic ketogenesis.30 Thereafter, from 12 h of age, healthy term infants show high ketone body turnover rates (12e 22 mmol kg/min) approaching those found in adults after several days of fasting,14 and during days 2 and 3 after birth they exhibit high ketone body concentrations quantitatively similar to those observed after an overnight fast in older children (Fig. 2).15 Such ketone body concentrations may account for as much as 25% of the neonate’s basal energy requirements during this time. Thus vigorous ketogenesis appears to be an integral part of extrauterine metabolic adaptation in the term human neonate.”

[4] Evidence type: review of experiments in humans and rats

Medina JM, Tabernero A.
J Neurosci Res. 2005 Jan 1-15;79(1-2):2-10.

“Although the supply of metabolic substrates is maintained mostly during the perinatal period, there is an apparent lack of mobilization of energy reserves immediately after delivery; i.e., during the presuckling period. During this period, the maternal supply of glucose has ceased and alternative substrates have not yet been released. In the rat, fatty acids come exclusively from the mother’s milk because of the lack of white adipose tissue at birth. Consequently, free fatty acids are not available in the rat before the onset of suckling (Mayor and Cuezva, 1985; Girard, 1990). In the case of human newborns, however, fatty acid mobilization occurs immediately after birth, although the onset of ketogenesis is delayed, probably as a consequence of a limited supply of carnitine, which is provided mainly by the milk (Hahn and Novak, 1985; Schmidt-Sommerfeld and Penn, 1990). In addition, glycogenolysis and gluconeogenesis are not active immediately after birth, resulting in very low concentrations of plasma glucose (Mayor and Cuezva, 1985; Girard, 1990). In these circumstances, lactate may play an important role as an alternative substrate. In fact, lactate accumulates in fetal blood during the perinatal period and is removed rapidly immediately after delivery (Persson and Tunell, 1971; Juanes et al., 1986).”

[5] Evidence type: presumably this is a review.
We could not get the full text, so for us this is evidence by authority

Shambaugh GE 3rd.
Fed Proc. 1985 Apr;44(7):2347-51.

“Pregnancy is characterized by a rapid accumulation of lipid stores during the first half of gestation and a utilization of these stores during the latter half of gestation. Lipogenesis results from dietary intake, an exaggerated insulin response, and an intensified inhibition of glucagon release. Increasing levels of placental lactogen and a heightened response of adipose tissue to additional lipolytic hormones balance lipogenesis in the fed state. Maternal starvation in late gestation lowers insulin, and lipolysis supervenes. The continued glucose drain by the conceptus aids in converting the maternal liver to a ketogenic organ, and ketone bodies produced from incoming fatty acids are not only utilized by the mother but cross the placenta where they are utilized in several ways by the fetus: as a fuel in lieu of glucose; as an inhibitor of glucose and lactate oxidation with sparing of glucose for biosynthetic disposition; and for inhibition of branched-chain ketoacid oxidation, thereby maximizing formation of their parent amino acids. Ketone bodies are widely incorporated into several classes of lipids including structural lipids as well as lipids for energy stores in fetal tissues, and may inhibit protein catabolism. Finally, it has recently been shown that ketone bodies inhibit the de novo biosynthesis of pyrimidines in fetal rat brain slices. Thus during maternal starvation ketone bodies may maximize chances for survival both in utero and during neonatal life by restraining cell replication and sustaining protein and lipid stores in fetal tissues.”

[6] Evidence type:

Herrera, Emilio
Endocrine, Volume 19, Number 1, October 2002 , pp. 43-56(14)

“During early pregnancy there is an increase in body fat accumulation, associated with both hyperphagia and increased lipogenesis. During late pregnancy there is an accelerated breakdown of fat depots, which plays a key role in fetal development. Besides using placental transferred fatty acids, the fetus benefits from two other products: glycerol and ketone bodies. Although glycerol crosses the placenta in small proportions, it is a preferential substrate for maternal gluconeogenesis, and maternal glucose is quantitatively the main substrate crossing the placenta. Enhanced ketogenesis under fasting conditions and the easy transfer of ketones to the fetus allow maternal ketone bodies to reach the fetus, where they can be used as fuels for oxidative metabolism as well as lipogenic substrates.”
“Increased gluconeogenesis from glycerol and ketogenesis from NEFA may benefit the fetus, which at late gestation is at its maximum accretion rate and its requirements for substrates and metabolic fuels are greatly augmented. The preferential use of glycerol for gluconeogenesis and the efficient placental transfer of the newly formed glucose may be of major importance to the fetus under these fasting conditions (Fig. 2), in which the availability of other essential substrates such as amino acids is reduced (30,34). Placental transfer of ketone bodies is highly efficient (35), reaching fetal plasma at the same level as in maternal circulation (29). Ketone bodies may be used by the fetus as fuels (36) and as substrates for brain lipid synthesis (37).”

[7] Evidence type: review
We could not get the full text, so for us this is evidence by authority

Girard J.
Biol Neonate. 1986;50(5):237-58.

Birth in most mammalian species represents an abrupt change from a high-carbohydrate and low-fat diet to a high-fat and low-carbohydrate diet. Gluconeogenesis is absent from the liver of the fetus of well fed mothers, but can be induced prematurely by prolonged fasting of the mother. Gluconeogenesis increases rapidly in the liver of newborn mammals in parallel with the appearance of phosphoenolpyruvate carboxykinase (PEPCK), the rate-limiting enzyme of this pathway. The rise in plasma glucagon and the fall in plasma insulin which occur immediately after birth are the main determinants of liver PEPCK induction. When liver PEPCK has reached its adult value, i.e. 24 h after birth, other factors are involved in the regulation of hepatic gluconeogensis. In order to maintain a high gluconeogenic rate, the newborn liver must be supplied with sufficient amount of gluconeogenic substrates and free fatty acids. An active hepatic fatty acid oxidation is necessary to support hepatic gluconeogenesis by providing essential cofactors such as acetyl CoA and NADH. The relevance of animal studies for the understanding of neonatal glucose homeostasis in man is discussed.”

[8] Evidence type: review of experiments:

Kalhan S, Parimi P.
Semin Perinatol. 2000 Apr;24(2):94-106.

(emphasis ours)
“Studies in human and animal models have consistently confirmed the dependence of the fetus on the mother for supply of glucose so that the fetus in utero under normal physiological circumstances does not produce glucose. However, most gluconeogenic and glycogenolytic enzymes have been shown to be present early in fetal development. The exception is the cytosolic phosphoenol pyruvate carboxykinase, which is expressed (at least in the rat) immediately after birth. 12-14 The appearance of gluconeogenic enzyme activity in the liver in relation to birth in the rat fetus and newborn is displayed in Figure 2. As shown, PC and glucose-6-phosphatase activity are expressed in the fetus, are relatively low at birth, and increase rapidly thereafter. Fructose 1,6-diphosphatase activity increases before birth. In contrast, phosphoenol pyruvate carboxykinase activity is absent in the fetus and rapidly increases immediately after birth, so that hepatic gluconeogenesis is completely absent in utero and appears in the immediate newborn period 2,14,5 GNG, however, can,be induced in utero by prolonged maternal starvation, prolonged hypoglycemia in the mother, or by direct injection of cyclic adenosine monophosphate (cAMP) into the fetus. 16-18 In addition, some studies have showed incorporation of tracer carbon from lactate into glucose in rat fetus and glutamine carbon into hepatic glycogen in sheep fetus. 4,5,19 The significance of these latter observations remains unclear.”

[9] Evidence type: meta-analysis

Walfisch A, Sermer C, Cressman A, Koren G.
BMJ Open. 2013 Aug 23;3(8):e003259. doi: 10.1136/bmjopen-2013-003259.

“The association between breastfeeding and child cognitive development is conflicted by studies reporting positive and null effects. This relationship may be confounded by factors associated with breastfeeding, specifically maternal socioeconomic class and IQ.
Systematic review of the literature.
Setting and participants
Any prospective or retrospective study, in any language, evaluating the association between breastfeeding and cognitive development using a validated method in healthy term infants, children or adults, was included.
Primary and secondary outcome measures
Extracted data included the study design, target population and sample size, breastfeeding exposure, cognitive development assessment tool used and participants’ age, summary of the results prior to, and following, adjustment for confounders, and all confounders adjusted for. Study quality was assessed as well.
84 studies met our inclusion criteria (34 rated as high quality, 26 moderate and 24 low quality). Critical assessment of accepted studies revealed the following associations: 21 null, 28 positive, 18 null after adjusting for confounders and 17 positive—diminished after adjusting for confounders. Directionality of effect did not correlate with study quality; however, studies showing a decreased effect after multivariate analysis were of superior quality compared with other study groupings (14/17 high quality, 82%). Further, studies that showed null or diminished effect after multivariate analysis corrected for significantly more confounders (7.7±3.4) as compared with those that found no change following adjustment (5.6±4.5, p=0.04). The majority of included studies were carried out during childhood (75%) and set in high-income countries (85.5%).
Much of the reported effect of breastfeeding on child neurodevelopment is due to confounding. It is unlikely that additional work will change the current synthesis. Future studies should attempt to rigorously control for all important confounders. Alternatively, study designs using sibling cohorts discordant for breastfeeding may yield more robust conclusions.”

[10] Evidence type: observational

Florey CD, Leech AM, Blackhall A.
Int J Epidemiol. 1995;24 Suppl 1:S21-6.

To determine the relationship between type of infant feeding and mental and psychomotor development at age 18 months.
A follow-up study of children born to primigravidae living in Dundee and booked into antenatal clinics in the City of Dundee (Local Authority District) from 1 May 1985 to 30 April 1986. The study population was 846 first born singletons, of whom 592 attended for developmental assessment at age 18 months. The main outcome measures were the Bayley Scales of Infant Mental and Motor Development.
Higher mental development was significantly related to breast feeding on discharge from hospital and according to the health visitors’ notes at about 2 weeks after discharge after allowing for partner’s social class, mother’s education, height, alcohol and cigarette consumption; placental weight and the child’s sex, birth weight and gestational age at birth. After adjustment for statistically significant variables, the difference in Bayley mental development index between breast and bottle fed infants was between 3.7 and 5.7 units depending on the source of feeding data. No differences were found for psychomotor development or behaviour.
The study provides further evidence of a robust statistical association between type of feeding and child intelligence. However, the literature is replete with suggestions for potential confounding variables which offer alternative causal explanations. To unravel what is an important clinical and public health question, further research should concentrate on randomized trials of supplemented formula feeds for children of mothers opting for bottle feeding and on epidemiological studies designed to disentangle the relation between method of feeding, parental intelligence and social environment.”

[11] Evidence type: observational

Belfort MB, Rifas-Shiman SL, Kleinman KP, Guthrie LB, Bellinger DC, Taveras EM, Gillman MW, Oken E.
JAMA Pediatr. 2013 Sep;167(9):836-44.

“Breastfeeding may benefit child cognitive development, but few studies have quantified breastfeeding duration or exclusivity, nor has any study to date examined the role of maternal diet during lactation on child cognition.
To examine relationships of breastfeeding duration and exclusivity with child cognition at ages 3 and 7 years and to evaluate the extent to which maternal fish intake during lactation modifies associations of infant feeding with later cognition.
Prospective cohort study (Project Viva), a US prebirth cohort that enrolled mothers from April 22, 1999, to July 31, 2002, and followed up children to age 7 years, including 1312 Project Viva mothers and children.
Duration of any breastfeeding to age 12 months.
Child receptive language assessed with the Peabody Picture Vocabulary Test at age 3 years, Wide Range Assessment of Visual Motor Abilities at ages 3 and 7 years, and Kaufman Brief Intelligence Test and Wide Range Assessment of Memory and Learning at age 7 years.
Adjusting for sociodemographics, maternal intelligence, and home environment in linear regression, longer breastfeeding duration was associated with higher Peabody Picture Vocabulary Test score at age 3 years (0.21; 95% CI, 0.03-0.38 points per month breastfed) and with higher intelligence on the Kaufman Brief Intelligence Test at age 7 years (0.35; 0.16-0.53 verbal points per month breastfed; and 0.29; 0.05-0.54 nonverbal points per month breastfed). Breastfeeding duration was not associated with Wide Range Assessment of Memory and Learning scores. Beneficial effects of breastfeeding on the Wide Range Assessment of Visual Motor Abilities at age 3 years seemed greater for women who consumed 2 or more servings of fish per week (0.24; 0.00-0.47 points per month breastfed) compared with less than 2 servings of fish per week (−0.01; −0.22 to 0.20 points per month breastfed) (P = .16 for interaction).
Our results support a causal relationship of breastfeeding duration with receptive language and verbal and nonverbal intelligence later in life.”

[12] Evidence type: observational

Julvez J, Guxens M, Carsin AE, Forns J, Mendez M, Turner MC, Sunyer J.
Dev Med Child Neurol. 2014 Feb;56(2):148-56. doi: 10.1111/dmcn.12282. Epub 2013 Oct 1.

This study investigated whether duration of full breastfeeding is associated with child neuropsychological development and whether this association is explained by social, psychological, and nutritional factors within families.
Participants in this study were a population-based birth cohort in the city of Sabadell (Catalonia, Spain). Females were recruited during the first trimester of pregnancy between July 2004 and July 2006. Information about parental characteristics and breastfeeding was obtained through questionnaires. Full breastfeeding was categorized as never, short term (≤4mo), long term (4-6mo), or very long term (>6mo). A trained psychologist assessed the neuropsychological development of children at 4 years of age (n=434) using the McCarthy Scales of Children’s Abilities (MSCA).
Full breastfeeding showed an independent association with child general MSCA scores after adjusting for a range of social, psychological, and nutritional factors (>6mo, coefficient=7.4 [95% confidence interval=2.8-12.0], p=0.011). Maternal social class, education level, and IQ were also associated with child neuropsychological scores, but did not explain breastfeeding associations. Omega-3 (n3) fatty acid levels were not associated with child neuropsychological scores.
Very long-term full breastfeeding was independently associated with neuropsychological functions of children at 4 years of age. Maternal indicators of intelligence, psychopathology, and colostrum n3 fatty acids did not explain this association.”

[13] Evidence type: observational

I Kull, M Wickman, G Lilja, S Nordvall, and G Pershagen
Arch Dis Child. 2002 December; 87(6): 478–481. doi: 10.1136/adc.87.6.478 PMCID: PMC1755833

“Aims: To investigate the effect of breast feeding on allergic disease in infants up to 2 years of age.
Methods: A birth cohort of 4089 infants was followed prospectively in Stockholm, Sweden. Information about various exposures was obtained by parental questionnaires when the infants were 2 months old, and about allergic symptoms and feeding at 1 and 2 years of age. Duration of exclusive and partial breast feeding was assessed separately. Symptom related definitions of various allergic diseases were used. Odds ratios (OR) and 95% confidence intervals (CI) were estimated in a multiple logistic regression model. Adjustments were made for potential confounders.
Results: Children exclusively breast fed during four months or more exhibited less asthma (7.7% v 12%, ORadj = 0.7, 95% CI 0.5 to 0.8), less atopic dermatitis (24% v 27%, ORadj = 0.8, 95% CI 0.7 to 1.0), and less suspected allergic rhinitis (6.5% v 9%, ORadj = 0.7, 95% CI 0.5 to 1.0) by 2 years of age. There was a significant risk reduction for asthma related to partial breast feeding during six months or more (ORadj = 0.7, 95% CI 0.5 to 0.9). Three or more of five possible allergic disorders—asthma, suspected allergic rhinitis, atopic dermatitis, food allergy related symptoms, and suspected allergic respiratory symptoms after exposure to pets or pollen—were found in 6.5% of the children. Exclusive breast feeding prevented children from having multiple allergic disease (ORadj = 0.7, 95% CI 0.5 to 0.9) during the first two years of life.
Conclusion: Exclusive breast feeding seems to have a preventive effect on the early development of allergic disease—that is, asthma, atopic dermatitis, and suspected allergic rhinitis, up to 2 years of age. This protective effect was also evident for multiple allergic disease.”

[14] Evidence type: review of observational studies

Pereira PF, Alfenas Rde C, Araújo RM.
J Pediatr (Rio J). 2014 Jan-Feb;90(1):7-15. doi: 10.1016/j.jped.2013.02.024. Epub 2013 Oct 16.

the aim of this study was to perform a review to investigate the influence of breastfeeding as a protective agent against the onset of diabetes in children.
non-systematic review of SciELO, LILACS, MEDLINE, Scopus, and VHL databases, and selection of the 52 most relevant studies. A total of 21 articles, specifically on the topic, were analyzed (nine related to type 1 diabetes and 12 to type 2 diabetes).
Data synthesis
the duration and exclusivity of breastfeeding, as well as the early use of cow’s milk, have been shown to be important risk factors for developing diabetes. It is believed that human milk contains substances that promote the maturation of the immune system, which protect against the onset of type 1 diabetes. Moreover, human milk has bioactive substances that promote satiety and energy balance, preventing excess weight gain during childhood, thus protecting against the development of type 2 diabetes. Although the above mentioned benefits have not been observed by some researchers, inaccuracies on dietary habit reports during childhood and the presence of interfering factors have been considered responsible for the lack of identification of beneficial effects.
given the scientific evidence indicated in most published studies, it is believed that the lack of breastfeeding can be a modifiable risk factor for both type 1 and type 2 diabetes. Strategies aiming at the promotion and support of breastfeeding should be used by trained healthcare professionals in order to prevent the onset of diabetes.”

[15] Evidence type: observational

Sun Y, Vestergaard M, Christensen J, Olsen J.
J Pediatr. 2011 Jun;158(6):924-9. doi: 10.1016/j.jpeds.2010.11.035. Epub 2011 Jan 13.

We asked whether breastfeeding reduces the risk of epilepsy in childhood.
We included 69 750 singletons born between September 1997 and June 2003 in the Danish National Birth Cohort and observed them to August 2008. Information on breastfeeding was reported by mothers in two computer-assisted telephone interviews at 6 and 18 months after birth. Information on epilepsy (inpatients and outpatients) was retrieved from the Danish National Hospital Register. Cox proportional hazards regression models were used to estimate incidence rate ratios and 95% CIs.
Breastfeeding was associated with a decreased risk of epilepsy, with a dose-response like pattern. For example, children breastfed for 3 to 5, 6 to 8, 9 to 12, and ≥ 13 months had a 26%, 39%, 50%, and 59% lower risk of epilepsy after the first year of life, respectively, compared with children who were breastfed for <1 month. The association remained when we excluded children who had adverse neonatal conditions or children who were exposed to adverse maternal conditions during pregnancy.
The observed protective effect of breastfeeding may be causal. Breastfeeding may decrease epilepsy in childhood, thereby adding another reason for breastfeeding.”

[16] Evidence type: controlled human experiments

“Our summary statistic, median peak kb [(ketone body)] concentration (Table 6), is significantly higher in the BF [(breastfed)] group compared with other feed groups for the SGA [(small for gestational age)] infants analyzed separately. We further explored the relationship between the blood glucose concentration and kb response by finding the kb concentration at the lowest blood glucose level for each infant at >24 hours of age (Fig 3, Table 6). Especially at low blood glucose values, infants who receive breast milk show some of the highest values for blood kb concentration. Our data show that exclusive formula feeding does not necessarily protect against low blood glucose values. Hence, the SGA FF [(formula fed)] infant could be doubly at risk of both low blood glucose values with a reduced kb response. No BF infant had both low blood glucose and low kb levels. For LGA [(large for gestational age)] infants, low blood glucose values were offset by kb concentrations of the same order of magnitude previously demonstrated for AGA [(appropriate for getstaional age)] infants6 (Fig 3).”
“Mammalian animal studies have shown that the postnatal induction of the enzymes involved in β-oxidation within the mitochondria requires the presence of long-chain fatty acids.15 The carnitine palmitoyltransferase system, which controls movement of long-chain fatty acids into the mitochondria, represents a major rate-limiting step in ketogenesis in the suckling rat. Long-chain fatty acids play a pivotal role in the posttranscriptional regulation of carnitine palmitoyltransferase 1 during the immediate postnatal period. We speculate that a factor present in breast milk but absent in formula milk augments ketogenesis in human neonates in the same way. Carnitine is known to have a central role in β-oxidation of fats: it is responsible for the transport of fatty acyl-coenzyme A across the inner mitochondrial membrane.16 During the suckling period, the demand for carnitine exceeds the rate of endogenous synthesis by up to 50%.17 Indeed, healthy, full-term infants fed formulas devoid of carnitine showed reduction in ketogenesis and an accumulation of fatty acid precursors in the plasma. Although breast milk– and cow’s milk– derived formulas contain equivalent amounts of carnitine,18 it may well be that there are significant differences in bioavailability. When compared with breastfed control subjects, infants who were fed a standard formula that was not supplemented with carnitine demonstrated markers of carnitine deficiency.19 Furthermore, we hypothesized that high intakes of energy and protein associated with early formula feeding may “switch off” or dampen the crucial glucagon surge, central to regulation of fuel availability in the immediate postnatal period.”

Language Barriers: Preferred Fuel

by Amber Wilcox-O’Hearn
As a linguist, I am often struck by the way phrasing and word choices not only reflect biases, but can actually create them [1].
Again and again in my researches into dietary science, I come upon expressions and turns of phrase that make Low Carb, High Fat diets seem intuitively unhealthy or illogical.
One such phrase is “Preferred Fuel”.

What is a preferred fuel?

Technically, a preferred fuel is simply one that is used first when there is more than one type available.
For many (but not all) of the tissues and organs in the body, glucose is a “preferred fuel”, meaning that
if glucose is available, those tissues and organs will use it before using most other fuels, even if they are also present.
People sometimes take this to mean that glucose is the best fuel; the fuel that would make you most healthy.
Choosing and preferring are things people do when they have decided that something is “better” than something else.
So, whether or not the physiologist who coined the term intended it, this terminology leads the reader to think that the preferred fuel is the one that is inherently a better choice.
However, this is not correct.
Sometimes the opposite is true.
Sometimes it depends on what else is going on.
One analogy I like is that of spending.
Many people take a large portion of their hard-earned pay cheques and make a payment on their debts.
No matter what else they do, they do that first.
Does this mean it is what they “prefer” to spend their money on?
In some sense, yes!
By the Doctrine of Revealed Preference [2], it is by definition what they prefer.
And yet most people would not count this kind of obligation as one of their choices.
There is another phrase, “disposable income” for spending associated with choice.
The reason paying debts is preferred in this way, is that one seeks to avoid the consequences of not paying.
In fact, in most cases, for most people, paying debts is a better choice than not paying them.
What it does not mean, however, is that people prefer to have debts to pay in the first place!
To bring this back to the realm of human metabolism, there are several substances that can be used for fuel, including fat, protein, ketones, glucose (which is the kind of fuel carbohydrates provide), and alcohol.
Alcohol is a strongly preferred fuel [3].
It is metabolized right away, even if there are other fuel sources around.
This is not evidence that alcohol is the best fuel, but rather that it is what the body wants to get rid of immediately.
Getting rid of the alcohol is a better choice than leaving it around to do damage.
The same is often true of glucose.
High blood sugar is toxic, and so whenever there is excess available it will get used or stored as quickly as possible.
We don’t mean to imply that being used for fuel first always means a substance is toxic.
What is a good “choice” in a system as complex as the human body is not always obvious.
For example, the heart uses acetoacetate (a ketone body) before glucose, but uses primarily fat, even in glycolytic dieters [4].

What about the brain?

It is true that the brain has need for some amount of glucose, even under a ketogenic metabolism.
That is one reason why gluconeogenesis (the ongoing production of glucose in the liver) is important, whether you are ketogenic or not.
This ensures a steady supply of glucose into the bloodstream, and keeps blood sugar in healthy bounds: enough for the brain, but not too much.
Under ketogenic conditions, though, the brain’s need for glucose is drastically reduced, because other fuels, such as ketone bodies and lactate take the place of most of it [5].
Lactate may be preferred over glucose in the brain [6], [7].
Using lactate instead of glucose has neuroprotective properties not unlike those seen with ketosis [8].
It might even be the case that a primary use of glucose in the brain is to be turned into lactate by astrocytes [8], [9].
We still have much to learn.
Next time you see the phrase “Preferred Fuel”, remember that it really just means the order in which different fuels are consumed.
There are various context-specific reasons a particular fuel might be used before other ones, and it has nothing to do with its inherent healthiness.


[1] See for example:

Fausey, Caitlin M., and Lera Boroditsky.
Psychonomic bulletin & review 17.5 (2010): 644-650.

Evidence type: controlled randomised experiment
“When bad things happen, how do we decide who is to blame and how much they should be punished? In the present studies, we examined whether subtly different linguistic descriptions of accidents influence how much people blame and punish those involved. In three studies, participants judged how much people involved in particular accidents should be blamed and how much they should have to pay for the resulting damage. The language used to describe the accidents differed subtly across conditions: Either agentive (transitive) or non-agentive (intransitive) verb forms were used. Agentive descriptions led participants to attribute more blame and request higher financial penalties than did nonagentive descriptions. Further, linguistic framing influenced judgments, even when participants reasoned about a well-known event, such as the “wardrobe malfunction” of Super Bowl 2004. Importantly, this effect of language held, even when people were able to see a video of the event. These results demonstrate that even when people have rich established knowledge and visual informa – tion about events, linguistic framing can shape event construal, with important real-world consequences. Subtle differences in linguistic descriptions can change how people construe what happened, attribute blame, and dole out punishment.

[2] From Investopedia
“Definition of ‘Revealed Preference’
“An economic theory of consumption behavior which asserts that the best way to measure consumer preferences is to observe their purchasing behavior. Revealed preference theory works on the assumption that consumers have considered a set of alternatives before making a purchasing decision. Thus, given that a consumer chooses one option out of the set, this option must be the preferred option.
“Revealed preference theory was introduced by Paul Samuelson in 1938. Since then it has expanded upon by a number of economists and remains a major theory of consumption behavior. The theory is especially useful in providing a method for analyzing consumer choice empirically.”
[3] See for example:
Ethanol and lipids.
Lieber CS, Savolainen M.
Alcohol Clin Exp Res. 1984 Jul-Aug;8(4):409-23.
“The interaction of ethanol with lipid metabolism is complex. When ethanol is present, it becomes a preferred fuel for the liver and displaces fat as a source of energy.”
Ethanol Causes Acute Inhibition of Carbohydrate, Fat, and Protein Oxidation and Insulin Resistance.
John J. Shelmet, George A. Reichard, Charles L. Skutches, Robert D. Hoeldtke, Oliver E. Owen, and Guenther Boden
J Clin Invest. 1988 April; 81(4): 1137–1145.
“We conclude that ethanol was a preferred fuel preventing fat, and to lesser degrees, CHO and protein, from being oxidized. It also caused acute insulin resistance which was compensated for by hypersecretion of insulin.”
[4] Evidence type: authority

Biochemistry. 5th edition.
Berg JM, Tymoczko JL, Stryer L.
New York: W H Freeman; 2002.

“Unlike skeletal muscle, heart muscle functions almost exclusively aerobically, as evidenced by the density of mitochondria in heart muscle. Moreover, the heart has virtually no glycogen reserves. Fatty acids are the heart’s main source of fuel, although ketone bodies as well as lactate can serve as fuel for heart muscle. In fact, heart muscle consumes acetoacetate in preference to glucose.”
Public Note to Self: “functions almost exclusively aerobically, as evidenced by the density of mitochondria” is relevant to the idea that ketogenic diets increase mitochondrial number. There is preliminary evidence of this, as discussed previously. This statement leads me to believe that it should be unsurprising if decreased reliance on glucose had this effect.

[5] Evidence type: controlled experiment

O. E. Owen, A. P. Morgan, H. G. Kemp, J. M. Sullivan, M. G. Herrera, and G. F. Cahill, Jr.
J Clin Invest. 1967 October; 46(10): 1589–1595. doi: 10.1172/JCI105650 PMCID: PMC292907

(Measurements on three people after 38-41 days of fasting)
“Turning to Table V, the average glucose uptake, after subtracting the amount glycolyzed to lactate and pyruvate, is 0.145 mmole/liter (2.6 mg/100 ml). This is markedly less than the usual 9-10 mg/100 ml observed in our laboratory while the techniques which little or no production of lactate or pyruvate was observed (25-28). Similar data were observed in our laboratory while the techniques and methods used to study these fasted subjects were validated. With measured cerebral blood flow of 45 ml/100 g of tissue per min, and assuming a brain size of 1400 g the 24 hr glucose oxidation would approximate 24 g, which agrees well with the theoretical maximum of 33 g calculated from nitrogen execretion and glycerol from adipose tissue as described above. The third confirmatory evidence for this marked reduction in glucose metabolism has been data, which obtained from hepatic and renal vein catheterization studies, demonstrated that the liver almost totally ceases to synthesize glucose from amino acids and that the kidney assumes the role of the major source of this diminished amount of glucose daily produced and consumed during starvation.”

[6] Evidence type: non-human animal experiments

Matthias T. Wyss, Renaud Jolivet, Alfred Buck, Pierre J. Magistretti, and Bruno Weber
The Journal of Neuroscience, 18 May 2011, 31(20): 7477-7485; doi: 10.1523/JNEUROSCI.0415-11.2011

“Cerebral energy metabolism is a highly compartmentalized and complex process in which transcellular trafficking of metabolites plays a pivotal role. Over the past decade, a role for lactate in fueling the energetic requirements of neurons has emerged. Furthermore, a neuroprotective effect of lactate during hypoglycemia or cerebral ischemia has been reported. The majority of the current evidence concerning lactate metabolism at the cellular level is based on in vitro data; only a few recent in vivo results have demonstrated that the brain preferentially utilizes lactate over glucose. Using voltage-sensitive dye (VSD) imaging, beta-probe measurements of radiotracer kinetics, and brain activation by sensory stimulation in the anesthetized rat, we investigated several aspects of cerebral lactate metabolism. The present study is the first in vivo demonstration of the maintenance of neuronal activity in the presence of lactate as the primary energy source. The loss of the voltage-sensitive dye signal found during severe insulin-induced hypoglycemia is completely prevented by lactate infusion. Thus, lactate has a direct neuroprotective effect. Furthermore, we demonstrate that the brain readily oxidizes lactate in an activity-dependent manner. The washout of 1-[ 11C]L-lactate, reflecting cerebral lactate oxidation, was observed to increase during brain activation from 0.077 Ϯ 0.009 to 0.105 Ϯ 0.007 min Ϫ1. Finally, our data confirm that the brain prefers lactate over glucose as an energy substrate when both substrates are available. Using [18F]fluorodeoxyglucose (FDG) to measure the local cerebral metabolic rate of glucose, we demonstrated a lactate concentration-dependent reduction of cerebral glucose utilization during experimentally increased plasma lactate levels.”

[7] Evidence type: review of experiments

Medina JM, Tabernero A.
J Neurosci Res. 2005 Jan 1-15;79(1-2):2-10.

“Lactate metabolism is particularly relevant in the brain, in which lactate is preferred over glucose, glutamine, or ketone bodies (Arizmendi and Medina, 1983; Fernandez and Medina, 1986; Vicario et al., 1991).”
Note that the references in this passage are either in rat brains or in glucose-6-phosphatase deficient children, so the unqualified generalisation as stated here is not warranted.

[8] Evidence type: non-human animal experiments

Johanne E. Rinholm, Nicola B. Hamilton, Nicoletta Kessaris, William D. Richardson, Linda H. Bergersen, and David Attwell
J Neurosci. 2011 January 12; 31(2): 538–548. doi: 10.1523/JNEUROSCI.3516-10.2011

“In the brain’s grey matter, astrocytes have been suggested to export lactate (derived from glucose or glycogen) to neurons to power their mitochondria. In the white matter, lactate can support axon function in conditions of energy deprivation, but it is not known whether lactate acts by preserving energy levels in axons or in oligodendrocytes, the myelinating processes of which are damaged rapidly in low energy conditions. Studies of cultured cells suggest that oligodendrocytes are the cell type in the brain which consumes lactate at the highest rate, in part to produce membrane lipids presumably for myelin. Here we use pH imaging to show that oligodendrocytes in the white matter of the rat cerebellum and corpus callosum take up lactate via monocarboxylate transporters (MCTs), which we identify as MCT1 by confocal immunofluorescence and electron microscopy. Using cultured slices of developing cerebral cortex from mice in which oligodendrocyte lineage cells express GFP under the control of the Sox10 promoter, we show that a low glucose concentration reduces the number of oligodendrocyte lineage cells and myelination. Myelination is rescued when exogenous L-lactate is supplied. Thus, lactate can support oligodendrocyte development and myelination. In CNS diseases involving energy deprivation at times of myelination or remyelination, such as periventricular leukomalacia leading to cerebral palsy, stroke, and secondary ischaemia following spinal cord injury, lactate transporters in oligodendrocytes may play an important role in minimising the inhibition of myelination that occurs.”

[9] Evidence type: conceptual integration of experiments

Yudkoff M, Daikhin Y, Melø TM, Nissim I, Sonnewald U, Nissim I.
Annu Rev Nutr. 2007;27:415-30.

“The oxidation of glucose provides essentially all energy needed to maintain cerebral function. Glycolysis may be relatively more prominent in some cells or in specific sub-cellular compartments. Thus, the filopodia of astrocytes are too narrow to accommodate mitochondria, and these cells will activate glycolysis (and glycogenolysis) in order to provide the energy that maintains their vital function of removing from the synaptic cleft much of the glutamate and K+ that presynaptic neuronal terminals release upon depolarization (24, 46). The fate of the pyruvate generated via glycolysis remains a topic of active inquiry and debate. It may be that astrocytes do not immediately oxidize all pyruvate produced via glycolysis. Instead, they may convert some pyruvate to lactate and release the latter to the extracellular fluid, from which neurons extract it and oxidize it as a fuel. Neurons can respire on lactate (102), but they may require glucose as a substrate if they are to maintain large internal pools of glutamate and aspartate (125). Astrocytic release of lactate and sub-sequent neuronal oxidation may constitute a mechanism by which neuronal and metabolic activity are effectively coupled (33, 66). ”

How much protein is enough?

div.sidebarintext {
margin: 0 0 0.5em 1em ;
border: medium outset ;
padding: 1em ;
background-color: #ffffee ;
width: 40% ;
float: right ;
clear: right }

div.sidebarintext p.rubric {
font-family: sans-serif ;
font-size: medium }

It seems, from clinical claims and numerous anecdotes, that protein intake has to be below some threshold for ketogenesis to continue, all else being equal.
(Conditions are rarely equal: the effects of fat intake, calorie intake, the profile of amino acids in your diet, the type of fat in your diet, exercise, and frequency of eating also matter!)
It is commonly assumed that excess protein gets immediately turned into glucose by gluconeogenesis.
However, we’ve shown in a series of articles (part 1, part 2, part 3, part 4)
that such a mechanism is highly unlikely —
excess protein does not just get immediately turned into glucose.
The evidence points to gluconeogenesis being driven by demand for glucose, not supply of protein.
However, it does appear that above a certain level of protein intake ketogenesis declines.
So regardless of mechanism, as ketogenic dieters, we probably still need to limit protein.
It’s not clear how much is too much.
But how much is enough?
It is important not to turn a healthy, ketogenic diet into an unhealthy starvation diet!
In this article we review some answers to this question, and some unanswered questions.

In Brief

  • It’s important to get enough protein.
  • The RDA for protein is too low: if you are like most people, your health will suffer if you eat as little protein as the RDA requires.
  • Getting the minimum may not be optimal, but getting less than the minimum would be a mistake.
  • There are several different conditions that are commonly believed to affect protein requirements.
    In particular, exercise and weight loss have both been said to increase protein needs.
    We couldn’t find definitive support for either of those beliefs.
  • We are most interested in studies that apply to keto dieters.
    Evidence from experiments on the Protein-Sparing Modified Fast is quite likely to apply to keto dieters.
    However, experiments on Protein-Sparing Modified Fasts leave out an important factor: fat intake, which may reduce protein requirements by reducing the need to use protein for energy.
  • We estimate that females should get at least 1.2 g/kg of ideal weight per day, and males at least 1.4 g/kg of ideal weight. [Edit 2014-01-11: (this may possibly be lowered by fat intake; see below)]
Measuring protein intake: per body weight, not percent of calories
Let’s head off confusion by distinguishing different ways to measure protein intake.
You’ll sometimes see protein intake expressed in terms of percent of total calorie intake.
For example, the USDA recommends 10-35% of calories from protein.
But this measures how much you are eating, not how much you are using!
If you are losing weight, then (hopefully!) some of the calories you are burning are coming from your body fat instead of from your diet.
Measuring protein requirements as a percent of intake ignores those calories.
In this case, the percent of protein you are eating would be higher than the percent you are using.
A better way to express protein requirements is in terms of body weight.
You are made in large part out of protein and so the more of you there is, the more protein you need to maintain your structures.
We have seen protein requirements in the form of percent of your ideal body weight, percent of your lean mass, or simply percent of your body weight.
These make more sense than percent of calories in this context.

Protein is essential to the body

Protein is necessary for virtually every biological process.
Enzymes, antibodies, cellular receptors, and collagen are just a few examples of protein structures, but there are maybe 2 million different proteins functioning in the body [1].
Getting too little protein to provide for these structures could adversely affect many aspects of health, especially growth and maintenance of bone and muscle, and immune function [2], [3], [4].
Although the body recycles the amino acids that proteins are built from (this is called protein “turnover”), there are also constant losses that need to be replaced (“obligatory nitrogen losses”) [5].

The RDA is not enough

The lowest estimates of minimum requirements we have seen is the USDA RDA (it’s now called RDI) of 0.8g per kg of body weight per day (for adults) [6].
Even they admit, in their 1000+ page explanation of macronutrient RDAs (RDAs for protein, fat and carbohydrate) from the Institute of Medicine, that the method they used to determine adequate protein levels (called “nitrogen balance”) is likely to result in underestimates [7].
Because of this, and because there was a flaw in the mathematical model that was used to derive the recommendations, a 2010 study reanalysed the same data that was used to generate the USDA RDA recommendations.
They concluded that the recommendations should be 50% higher than they are [8].
That is, they conclude that for most people, 1.2g/kg/day of protein intake is necessary.
There is also experimental evidence that suggests the RDA is inadequate [9].

Different conditions may require different amounts of protein

An issue with using the results of these studies, is that, as with any physiological experiments, when you change the base conditions, the results may suddenly no longer apply.
Changes that might make these conclusions inapplicable include what happens during high levels of physical activity, weight loss, or when consuming different levels of macronutrients. Specifically we want to know if protein requirements change when keto-adapted.
Exercise conditions
Whether higher levels of exercise increase protein needs is controversial.
Some studies suggest that endurance exercise increases the need for protein because protein is oxidised (used for energy) during the exercise [10].
However, there is also evidence that after a period of training this oxidation is limited by downregulation of the necessary enzymes [11].
That is, after doing it regularly, the body learns to spare the protein and stops letting it be used in that way.
In particular, it is the BCAAs (branched chain amino acids) that appear to be oxidised [11].
Since BCAAs are known to be elevated in keto dieters,
it’s possible that any such effect would be minimal for us in the first place or that the downregulation has already taken place.
In fact, infusing people with beta-hydroxybutyrate (the ketone body we check for in the blood to see if someone is in deep ketosis) reduces leucine oxidation and promotes protein synthesis [12].

(Leucine is a BCAA — it is probably the most important BCAA for ketogenic metabolism. More on that another day.)

As for weight lifting,
it seems intuitive that if you are building muscle, you will need more protein.
Some studies seem to show this, but others actually show the reverse [13], [14], [15].
The subjects in those studies, of course, were not on keto diets.
These facts taken together seem to suggest that when you adapt to a hormonal state that favours muscle preservation or gain, whether it is a ketogenic state, or the state of regular exercise stimulus, it reduces the need for protein.
Indeed, ketosis is known to be protein-sparing, and that may mean that you actually need less protein.
Protein requirements and weight loss
We have seen it asserted that weight loss itself increases the need for protein.
However, the only studies we have seen that claim to prove it are flawed [16], [17].
It is still possible that weight loss requires more protein, but we haven’t yet been convinced either way.
Note that this is different from the claim that higher protein diets improve weight loss (even when carbohydrates are not ketogenically low), which has shown repeatedly [18], [19], [20], [21], [22].
Since protein is generally taking the place of carbohydrate, we would certainly expect that to be true.
Relatedly, it is often claimed that weight loss is necessarily accompanied by some loss of lean mass.
This has a plausible mechanism.
For one thing, in people with lots of excess fat, more muscle, bone, and other infrastrucure has been built to carry the excess weight, and as the stimulus of the weight of the fat recedes, the extra muscle built to hold it might atrophy.
Nonetheless, there are some studies of keto diets used for weight loss in which lean mass was preserved or even increased, for example in [24] and [25].

Protein requirements for keto dieters

Since retaining lean mass requires sufficient protein (among other things), studies in which keto dieters retained lean mass must have been protein adequate.
Note that the reverse is not true: if a study had keto dieters losing lean mass, that doesn’t imply that protein intake was inadequate.
Other factors, such as mineral intake [23], and activity level and type [24] can effect muscle retention.
The Protein-Sparing Modified Fast
One particular kind of ketogenic weight loss diet that specifically aims to provide just enough protein to prevent muscle loss is called a Protein-Sparing Modified Fast.
The developers of the Protein-Sparing Modified Fast were hypothesising that fasting would be the perfect therapy for the treatment of obesity, if it weren’t for the resulting detrimental and dangerous loss of lean mass.
They discovered that by adding enough protein, subjects lost fat effectively, but without losing lean mass.
In particular, it was found that males preserved lean mass at 1.4 g/kg ideal body weight, and females at 1.2 g/kg ideal body weight [25].
This is particularly relevant for us, since the Protein-Sparing Modified Fast is a ketogenic condition.
Adding fat into the mix
In addition to the effects of being keto-adapted, we would like to know what the effect of the amount of fat in the diet has, if any, on protein requirements.
It is plausible that if for some reason the rate of oxidation of fat stores were limited, that some dietary protein might be oxidised for fuel.
If this were the case, then the addition of fat should lessen protein requirements.
In other words, if there is a rate limit to getting fat out of fat stores, then adding sufficient dietary fat to make up for the energy requirements not met by adipose tissue would spare protein.
On a keto diet, since carbohydrates are limited, and protein levels that are too high are undesirable (though how high is “too high” is not well understood), energy balance is made up by dietary fat.
(And besides, fat is an important source of essential fatty acids and nutrients.)
Unfortunately, we don’t know of studies that test this in either weight loss or weight maintenance conditions.


It is important to get enough protein.
We looked at the RDA, and that is certainly too low, so we looked for other sources of evidence.
We also tried to determine if some common conditions would alter the amount needed, without coming to any strong conclusions.
We wanted to find evidence that would be relevant to ketogenic dieters in particular.
Studies based on the Protein-Sparing Modified Fast don’t match our needs exactly, since we eat lots of fat, but at least the participants were in ketosis.
Based on those studies, we think females need at least 1.2 g/kg of ideal body weight, and men at least 1.4.
However, we still don’t know how high protein can be before it interferes with ketosis, and this is likely to vary by individual.
Moreover, we don’t know if it would be better for your health to have just the minimum, or to have substantially more.
It is certainly plausible that having more would allow for better functioning.


[1] Evidence type: authority
“The word “proteome” is derived from PROTEins expressed by a genOME, and it refers to all the proteins produced by an organism, much like the genome is the entire set of genes. The human body may contain more than 2 million different proteins, each having different functions. As the main components of the physiological pathways of the cells, proteins serve vital functions in the body such as:

  • catalyzing various biochemical reactions, e.g. enzymes;
  • acting as messengers, e.g. neurotransmitters;
  • acting as control elements that regulate cell reproduction;
  • influencing growth and development of various tissues, e.g. trophic factors;
  • transporting oxygen in the blood, e.g. hemoglobin; and
  • defending the body against disease, e.g. antibodies.
[2] Evidence type: experimental

Castaneda C, Charnley JM, Evans WJ, Crim MC.
Am J Clin Nutr. 1995 Jul;62(1):30-9.

“A 9-wk study of adaptation to marginal protein intakes was conducted in 12 elderly women. Subjects were randomly assigned to two groups fed a weight-maintenance diet containing either 1.47 (low) or 2.94 (adequate) g body cell mass-1.d-1 (0.45 and 0.92 body wt-1.d-1, respectively). Mean nitrogen balance in the low-protein group remained negative throughout the study. These subjects experienced significant losses in lean tissue, immune response, and muscle function. The adequate-protein group was in nitrogen balance throughout the study, without changes in lean tissue, and with improvements in immune response, serum immunoglobulins, albumin, total protein values, and muscle function. Thus, elderly women fed the low-protein diet accommodated to the diet by compromising functional capacity, whereas those fed the adequate diet maintained functional capacity.”

[3] Evidence type: summary of experiments

Peng Li, Yu-Long Yin, Defa Li, Sung Woo Kim and Guoyao Wu.
British Journal of Nutrition (2007), 98, 237–252 doi: 10.1017/S000711450769936X

“A deficiency of dietary protein or amino acids has long been known to impair immune function and increase the susceptibility of animals and humans to infectious disease. However, only in the past 15 years have the underlying cellular and molecular mechanisms begun to unfold. Protein malnutrition reduces concentrations of most amino acids in plasma. Findings from recent studies indicate an important role for amino acids in immune responses by regulating: (1) the activation of T lymphocytes, B lymphocytes, natural killer cells and macrophages; (2) cellular redox state, gene expression and lymphocyte proliferation; and (3) the production of antibodies, cytokines and other cytotoxic substances. Increasing evidence shows that dietary supplementation of specific amino acids to animals and humans with malnutrition and infectious disease enhances the immune status, thereby reducing morbidity and mortality.”

[4] Evidence type: observational

Bonjour JP, Ammann P, Chevalley T, Rizzoli R.
Can J Appl Physiol. 2001;26 Suppl:S153-66.

“Among osteotrophic nutrients, proteins play an important role in bone development, thereby influencing peak bone mass. Consequently, protein malnutrition during development can increase the risk of osteoporosis and of fragility fracture later in life. Both animal and human studies indicate that low protein intake can be detrimental for both the acquisition of bone mass during growth and its conservation during adulthood. Low protein intake impairs both the production and action of IGF-I (Insulin-like growth factor-I). IGF-I is an essential factor for bone longitudinal growth, as it stimulates proliferation and differentiation of chondrocytes in the epiphyseal plate, and also for bone formation. It can be considered as a key factor in the adjustments of calcium-phosphate metabolism required for normal skeletal development and bone mineralization during growth. In healthy children and adolescents, a positive association between the amount of ingested proteins and bone mass gain was observed in both sexes at the level of the lumbar spine, the proximal femur and the midfemoral shaft. This association appears to be particularly significant in prepubertal children. This suggests that, like for the bone response to either the intake of calcium or weight-bearing exercise, the skeleton would be particularly responsive to the protein intake during the years preceding the onset of pubertal maturation.”

[5] Evidence type: summary of experiments

“As protein deficiency declines, the efficiency with which amino acids are reutilized increases, but small amounts of amino acids are continuously degraded, and their nitrogen is lost from the body even when no protein is being consumed. These losses are termed obligatory nitrogen losses. If protein intake declines below the amount of these obligatory losses, the amount of nitrogen excreted by the body will exceed the amount consumed, and the body will be in a state of negative nitrogen balance. As protein intake is increased, nitrogen balance will become less negative until, at some point, a steady state will be reached such that nitrogen intake will equal nitrogen loss; i.,e., nitrogen balance will be zero. An estimate of adult requirements for protein or amino acids can therefore be obtained from measurement of the minimum amount of protein or amino acid nitrogen that must be consumed to just balance body nitrogen losses.”

[6] Dietary Reference Intakes: Macronutrients

[7] Evidence type: summary of experiments

“The nitrogen balance method does have substantial practical limitations and problems. First, the rate of urea turnover in adults is slow, so several days of adaptation are required for each level of dietary protein tested to attain a new steady state of nitrogen excretion (Meakins and Jackson, 1996; Rand et al., 1976). Second, the execution of accurate nitrogen balance measurements requires very careful attention to all the details of the procedures involved. Since it is easy to overestimate intake and underestimate excretion, falsely positive nitrogen balances may be obtained (Hegsted, 1976). Indeed, an overestimate of nitrogen balance seems consistent throughout the literature because there are many observations of quite considerable apparent retention of nitrogen in adults (Oddoye and Margen, 1979). This observation is biologically implausible because (a) adults do not normally accrete body protein, and (b) the magnitude of the positive nitrogen balances is inconsistent with stability of body weight.”
Comment: That last statement is odd. It could be seen as denying the evidence. We actually find it plausible that adults could accrete protein to a greater extent than currently assumed (more on this in a subsequent post). And of course, as every dieter knows, body weight can be stable even while lean mass is increasing, if fat is simultaneously decreasing.

[8] Evidence type: meta-analysis of experiments

Elango R, Humayun MA, Ball RO, Pencharz PB.
Curr Opin Clin Nutr Metab Care. 2010 Jan;13(1):52-7.

“This review discusses recent evidence that suggests a significant underestimation of protein requirements in adult humans.
Traditionally, total protein requirements for humans have been determined using nitrogen balance. The recent Dietary Reference Intake recommendations for mean and population-safe intakes of 0.66 and 0.8 g/kg/day, respectively, of high-quality protein in adult humans are based on a meta-analysis of nitrogen balance studies using single linear regression analysis. We reanalyzed existing nitrogen balance studies using two-phase linear regression analysis and obtained mean and safe protein requirements of 0.91 and 0.99 g/kg/day, respectively. The two-phase linear regression analysis is considered more appropriate for biological analysis of dose-response curves. Considering the inherent problems associated with the nitrogen balance method, we developed an alternative method, the indicator amino acid oxidation technique, to determine protein requirements The mean and population-safe requirements in adult men were determined to be 0.93 and 1.2 g/kg/day and are 41 and 50%, respectively, higher than the current Dietary Reference Intakes recommendations.
The indicator amino acid oxidation-based requirement values of 0.93 and 1.2 g protein/kg/day and the reanalysis of existing nitrogen balance studies are significantly higher than current recommendations. Therefore, there is an urgent need to reassess recommendations for protein intake in adult humans.”

[9] Evidence type: experimental

Anna E. Thalacker-Mercer,1 James C. Fleet,1 Bruce A. Craig,2 and Wayne W. Campbell
J Nutr Biochem. 2010 November; 21(11): 1076–1082.

“Previous studies have suggested that the current RDA for protein (0.8 g•kg−1•d−1) is inadequate to maintain protein homeostasis [5, 32-34], fat-free mass, and skeletal muscle cross-sectional area in older adults [7, 35] while protein intake above the RDA may be needed to maintain lean body mass and type I muscle fiber cross sectional area in older women [5, 6]. The results of our study reveal a molecular signature of a biological response to inadequate dietary protein intake that is consistent with this hypothesis. We believe that the observation that transcripts for “muscle and organ development” and “cell differentiation” increase when the protein intake is greater than the RDA (i.e. cluster C5 in Figure 1) and previous phenotypic data [5-7, 32-35] suggest that there are positive responses to consuming protein in quantities greater than the RDA in skeletal muscle. Thus, our array data are consistent with previous reports that suggest a higher protein intake contributes to a healthier physiological profile [5, 6, 36] and better muscle responses to resistance training [37]. The study duration was not, however, sufficient to directly link changes in the skeletal muscle transcript profile with previously reported phenotypes of adaptation and accommodation [5-7, 32-35].”

[10] Evidence type: experiment

Lemon PW, Dolny DG, Yarasheski KE.
Can J Appl Physiol. 1997 Oct;22(5):494-503.

“Six healthy men completed three 1-hr bouts of treadmill walk-jogging at low (L; 42 +/- 3.9% VO2max), moderate (M; 55 +/- 5.6%), and high (H; 67 +/- 4.5%) exercise intensity in order to determine whether moderate physical activity affects dietary protein needs. Both sweat rate and sweat urea N loss were greater (p < .10) with increasing exercise intensity. Seventy-two hour postexercise urine urea N excretion was elevated (p < .05) over nonexercise control (26.6 +/- 2.96 g) with both M (31.0 +/- 3.65) and H (33.6 +/- 4.39), but not L (26.3 +/- 1.86), intensities. Total 72-hr postexercise urea N excretion (urine + sweat) for the M and H exercise was greater than control by 4.6 and 7.2 g, respectively. This suggests that 1 hr of moderate exercise increases protein oxidation by about 29-45 g, representing approximately 16-25% of the current North American recommendations for daily protein intake. These data indicate that the type of exercise typically recommended for health/wellness can increase daily protein needs relative either to sedentary individuals or to those who exercise at lower intensities.”

[11] Evidence type: review of experiments

Tarnopolsky M.
Nutrition. 2004 Jul-Aug;20(7-8):662-8.

“Human skeletal muscle can oxidize at least eight amino acids (alanine, asparagine, aspartate, glutamate, isoleucine, leucine, lysine, and valine; 70); however, during exercise, the branched-chain amino acids (BCAA: isoleucine, leucine, and valine) are preferentially oxidized (43, 49, 58, 70). The BCAA are transaminated to their keto-acids via branched-chain aminotransferase (BCAAT), with subsequent oxidation occurring via branched-chain oxo-acid dehydrogenase enzyme (BCOAD; 8, 9). The amino-N group is usually transaminated with α-ketoglutarate to form glutamate, which is then transaminated with pyruvate to form alanine or aminated via glutamine synthase to form glutamine (15, 57). The BCOAD enzyme is rate limiting in BCAA oxidation, with about 5 to 8% being active (dephosphorylated) at rest, and 20 to 25% being active during exercise (8, 49).”

[12] (Thank you to Bill Lagakos for pointing us to this study.)
Evidence type:

Nair KS, Welle SL, Halliday D, Campbell RG.
J Clin Invest. 1988 Jul;82(1):198-205.

“Because intravenous infusion of beta-hydroxybutyrate (beta-OHB) has been reported to decrease urinary nitrogen excretion, we investigated in vivo metabolism of leucine, an essential amino acid, using L-[1-13C]leucine as a tracer during beta-OHB infusion. Leucine flux during beta-OHB infusion did not differ from leucine flux during normal saline infusion in nine normal subjects, whereas leucine oxidation decreased 18-41% (mean = 30%) from 18.1 +/- 1.1 (P less than 0.01), and incorporation of leucine into skeletal muscle protein increased 5-17% (mean = 10%) from 0.048 + 0.003%/h (P less than 0.02). Since blood pH during beta-OHB infusion was higher than the pH during saline infusion, we performed separate experiments to study the effect of increased blood pH on leucine kinetics by infusing sodium bicarbonate intravenously. Blood pH during sodium bicarbonate infusion was similar to that observed during the beta-OHB infusion, but bicarbonate infusion had no effect on leucine flux or leucine oxidation. We conclude that beta-OHB decreases leucine oxidation and promotes protein synthesis in human beings.”

[13] Evidence type: uncontrolled experiment

Pikosky M, Faigenbaum A, Westcott W, Rodriguez N.
Med Sci Sports Exerc. 2002 May;34(5):820-7.

“Healthy children (N = 11, 8.6 +/- 1.1 yr, 33.7 +/- 9.4 kg, 131 +/- 9.6 cm, BMI = 19.1 +/- 3.4) participated in a supervised resistance-training program 2 times.wk-1 for 6 wk. 15N glycine methodology was used to assess nitrogen flux (Q), protein synthesis (PS), protein breakdown (PB), and net turnover ([NET] = PS – PB) before (PRE) and after (POST) resistance training. Percent body fat (%BF), fat-free mass (FFM), fat mass (FM), and energy and protein intakes were also determined. PRE/POST measurements of 1RM for the chest press and leg extension were used to examine strength gains.
Gains associated with the chest press and leg extension were 10% and 75% (P < 0.001), respectively. Significant increases (P < 0.05) were noted for weight, height, FFM, and FM. Energy and protein intake remained constant. Significant decreases (PRE vs POST) were observed for Q (1.22 +/- 0.1 vs 0.75 +/- 0.05, P < 0.001), PS (6.48 +/- 0.47 vs 3.55 +/- 0.30, P < 0.001), and PB (5.24 +/- 0.41 vs 2.96 +/- 0.30, P < 0.01) after 6 wk of resistance training. NET was also reduced (P = 0.07, 1.24 +/- 0.31 vs 0.59 +/- 0.20
Resistance training resulted in a downregulation in protein metabolism, which may be energy based. Future studies are needed to clarify energy, as well as protein, needs in young children participating in this form of exercise.”

[14] Evidence type: uncontrolled experiment

Hartman JW, Moore DR, Phillips SM.
Appl Physiol Nutr Metab. 2006 Oct;31(5):557-64.

“It is thought that resistance exercise results in an increased need for dietary protein; however, data also exists to support the opposite conclusion. The purpose of this study was to determine the impact of resistance exercise training on protein metabolism in novices with the hypothesis that resistance training would reduce protein turnover and improve whole-body protein retention. Healthy males (n = 8, 22 +/- 1 y, BMI = 25.3 +/- 1.8 kg.m(-2)) participated in a progressive whole-body split routine resistance-training program 5d/week for 12 weeks. Before (PRE) and after (POST) the training, oral [15N]-glycine ingestion was used to assess nitrogen flux (Q), protein synthesis (PS), protein breakdown (PB), and net protein balance (NPB = PS-PB). Macronutrient intake was controlled over a 5d period PRE and POST, while estimates of protein turnover and urinary nitrogen balance (N(bal) = N(in) – urine N(out)) were conducted. Bench press and leg press increased 40% and 50%, respectively (p < 0.01). Fat- and bone-free mass (i.e., lean muscle mass) increased from PRE to POST (2.5 +/- 0.8 kg, p < 0.05). Significant PRE to POST decreases (p <0.05) occurred in Q (0.9 +/- 0.1 vs. 0.6 +/- 0.1 g, PS (4.6 +/- 0.7 vs. 2.9 +/- 0.3, and PB (4.3 +/- 0.7 vs. 2.4 +/- 0.2 Significant training-induced increases in both NPB (PRE = 0.22 +/- 0.13; POST = 0.54 +/- 0.08 and urinary nitrogen balance (PRE = 2.8 +/- 1.7 g N.d(-1); POST = 6.5 +/- 0.9 g N.d(-1)) were observed. A program of resistance training that induced significant muscle hypertrophy resulted in reductions of both whole-body PS and PB, but an improved NPB, which favoured the accretion of skeletal muscle protein. Urinary nitrogen balance increased after training. The reduction in PS and PB and a higher NPB in combination with an increased nitrogen balance after training suggest that dietary requirements for protein in novice resistance-trained athletes are not higher, but lower, after resistance training.”

[15] Evidence type: uncontrolled experiment

Moore DR, Del Bel NC, Nizi KI, Hartman JW, Tang JE, Armstrong D, Phillips SM.
J Nutr. 2007 Apr;137(4):985-91.

“We aimed to determine the impact of intense resistance training, designed to increase lean body mass (LBM), on both fasted and fed whole body protein kinetics in untrained young men. Twelve healthy males (22 +/- 2 y of age; BMI, 24.3 +/- 2.4 kg/m(2)) participated in a 12-wk (5-d/wk) resistance training program. Before and after training, a primed constant infusion of [1-(13)C]leucine was used to measure whole body leucine turnover, protein breakdown, and nonoxidative leucine disposal in the fasted and fed states. Participants were studied during 5-d controlled diet periods that provided a moderate protein intake [1.4 g/(kg body wt . d)]. We estimated protein turnover and nitrogen balance. Training increased LBM (61.6 +/- 6.9 vs. 64.8 +/- 6.7 kg, P < 0.05). After training, whole body leucine turnover was reduced (P < 0.01) in both fasted (167 +/- 18 vs. 152 +/- 17) and fed (197 +/- 23 vs. 178 +/- 21) states [all values micromol/(kg LBM . h)]. Training-induced decreases (P < 0.01) in protein breakdown occurred in the fasted (165 +/- 18 vs. 144 +/- 17) and fed (111 +/- 23 vs. 93 +/- 20) states. Following training, nonoxidative leucine disposal was similarly reduced (P < 0.01) in the fasted (144 +/- 18 vs. 126 +/- 18) and fed (151 +/- 20 vs. 133 +/- 19) states. Nitrogen balance was more positive after training (13.7 +/- 8.1 vs. 33.4 +/- 12.5 g/(kg LBM . d), P < 0.01) indicating an increased retention of dietary nitrogen. Intense resistance training alters whole body protein kinetics in novice weightlifters regardless of feeding status. The increase in nitrogen balance after training demonstrates a more efficient utilization of dietary nitrogen, suggesting that protein requirements for novice weightlifters are not elevated.”

[16] Evidence type: experimental

L J Hoffer, B R Bistrian, V R Young, G L Blackburn, and D E Matthews.
J Clin Invest. 1984 March; 73(3): 750–758.

“A randomized comparison trial of two very low calorie weight reduction diets was carried out for 5 or 8 wk in 17 healthy obese women. One diet provided 1.5 g protein/kg ideal body weight; the other provided 0.8 g protein/kg ideal body weight plus 0.7 g carbohydrate/kg ideal body weight. The diets were isocaloric (500 kcal). Amino acid metabolism was studied by means of tracer infusions of L-[1-13C]leucine and L-[15N]alanine. After 3 wk of adaptation to the diets, nitrogen balance was zero for the 1.5 g protein diet but -2 g N/d for the 0.8 g protein diet. Postabsorptive plasma leucine and alanine flux decreased from base line by an equal extent with both diets by approximately 20 and 40%, respectively. It was concluded that protein intakes at the level of the recommended dietary allowance (0.8 g/kg) are not compatible with nitrogen equilibrium when the energy intake is severely restricted, and that nitrogen balance is improved by increasing the protein intake above that level. Basal rates of whole body nitrogen turnover are relatively well maintained, compared with total fasting, at both protein intakes. However, turnover in the peripheral compartment, as evidenced by alanine flux, may be markedly diminished with either diet.”
Comment: The authors have assumed that the RDA is adequate, which would imply that the condition of weight loss had increased protein needs. However, since we have seen that the RDA is not, in fact adequate, the conclusion is unwarranted. I.e., the study was not controlled for people not losing weight.

[17] Evidence type: experimental

W J Smith, L E Underwood and D R Clemmons.
The Journal of Clinical Endocrinology & Metabolism February 1, 1995 vol. 80 no. 2 443-449

“Sixteen adult subjects (7 males and 9 females), aged 22-40 yr, were recruited and assessed in the same manner as the children [i.e. by advertisement, and determined to be healthy by a battery of tests.] They consumed a diet that provided 35 Cal/kg IBW and 1 g protein/kg IBW for the first 3 study days. Nonprotein calories were provided as 70% carbohydrate and 30% lipid. On study days 4-9, half of the 16 subjects received only 17 Cal/kg IBW, but continued to receive 1 g protein/kg IBW.
“During the 6 days of calorie restriction, the adults lost 1.7 2 0.1 kg (mean 2 SD) […]. Baseline nitrogen balance in the adults (mean of days 2 and 3) was -82 2 47 mmol/day (mean + SE) and declined to -196 2 40 mmol/day (the mean of study days 8 and 9; P < 0.05).”
Comment: The authors showed that when healthy adults were given a constant amount of protein, but two different levels of calories, nitrogen balance was more negative in the lower calorie condition. However, both diets were still high enough in carbohydrate to be non-ketogenic, and the difference between the diets was primarily a carbohydrate difference. In a non-keto-adapted state, carbohydrates can spare protein, because your metabolism is much more dependent on glucose for fuel. That means that reducing carbohydrates will require more protein for gluconeogenesis. After keto-adaptation, many tissues use ketones instead, and this spares protein. So we would expect that lowering carbohydrates without keto-adapting would negatively affect nitrogen balance. Therefore, we can’t conclude that the negative nitrogen balance in this study reflects increased protein needs from caloric restriction alone, because it could be that the same caloric deficit in a keto-adapted state would have different results. It would be interesting to see an analogue of this study in keto-adapted people. We are not aware of any such study.

[18] Evidence type: randomised controlled trial

Baba NH, Sawaya S, Torbay N, Habbal Z, Azar S, Hashim SA.
Int J Obes Relat Metab Disord. 1999 Nov;23(11):1202-6.

“Thirteen male obese hyperinsulinemic normoglycemic subjects were divided into two groups and fed hypoenergetic diets providing 80% of their resting energy expenditure (REE). One group received a high-protein diet (HP; 45% protein, 25% carbohydrates, and 30% fat as percent of dietary energy) and the other a high-carbohydrate diet (HC; 12% protein, 58% carbohydrates and 30% fat).
“Weight loss was higher in the HP than HC group (8.3+/-0.7 vs 6.0+/-0.6 kg, P<0. 05). There was a decrease in body fat in both groups, whereas body water decreased significantly more in the HP group. REE decreased more in the HC than the HP group (-384.3+/-84.6 vs -132.3+/-51.0 kcal, P<0.05). Serum total cholesterol, triglycerides and LDL cholesterol decreased significantly to a similar extent in both diet groups, while HDL cholesterol was decreased significantly only in the HP group. Mean fasting insulin decreased significantly in both diet groups and reached the normal range only in the HP group.”

[19] Evidence type: randomized controlled trial

Labayen I, Díez N, González A, Parra D, Martínez JA.
Forum Nutr. 2003;56:168-70.

Eleven obese (BMI>30) women were randomly assigned to a 10w dietary intervention study comparing HP (30% protein) or HC (55% carbohydrate) energy restricted diets providing 30% energy fat content. Substrate utilisation was evaluated by indirect calorimetry. Body weight and composition (Bod Pod) and blood measurements were performed before and after weight loss.
On average, the individuals on the HP dietary group lost 4.4 kg more than those in the HC program (p<0.50), which was mainly due to a fat mass loss (3.7 kg, p<0.05) with no statistical differences in lean body mass reduction. These losses were accompanied by a significant decrease in fasting leptin in the HP group (-52%; P<0.05). On the other hand post-absorptive lipid oxidation decreased in the HC group (-48%) and remained unchanged in the HP groups.
The replacement of some dietary carbohydrate by protein in energy restricted diets, improves weight and fat losses and specifically promotes lipid oxidation in the fasting state, without major different in lean body mass depletion.”

[20] Evidence type: randomized controlled trial

Evangelista LS, Heber D, Li Z, Bowerman S, Hamilton MA, Fonarow GC.
J Cardiovasc Nurs. 2009 May-Jun;24(3):207-15. doi: 10.1097/JCN.0b013e31819846b9.

“[T]his feasibility study was conducted to compare the effects of an HP, hypoenergetic diet (40% total energy from carbohydrates, 30% from protein, and 30% from fat), a standard protein (SP), hypoenergetic diet (55% total energy from carbohydrates, 15% from protein, and 30% from fat), and a conventional diet (high carbohydrates, low fat, high fiber, with no energy restrictions) on body weight and adiposity in patients with HF and increased BMI greater than or equal to 27 kg/m2 and DM.
“The current report reflects data from the first 14 patients who consented to participate in the study; recruitment and enrollment are still ongoing. The patients were randomized to the HP (n = 5), SP (n = 5), and conventional diets (n = 4), respectively.
[…] Gee, look at that cholesterol. Not quite statistically significant (because the group size was small) but highly suggestive.

[21] Evidence type: experimental

Pasiakos SM, Cao JJ, Margolis LM, Sauter ER, Whigham LD, McClung JP, Rood JC, Carbone JW, Combs GF Jr, Young AJ.
FASEB J. 2013 Sep;27(9):3837-47. doi: 10.1096/fj.13-230227.

“The purpose of this work was to determine the effects of varying levels of dietary protein on body composition and muscle protein synthesis during energy deficit (ED). A randomized controlled trial of 39 adults assigned the subjects diets providing protein at 0.8 (recommended dietary allowance; RDA), 1.6 (2×-RDA), and 2.4 (3×-RDA) g kg(-1) d(-1) for 31 d. A 10-d weight-maintenance (WM) period was followed by a 21 d, 40% ED. Body composition and postabsorptive and postprandial muscle protein synthesis were assessed during WM (d 9-10) and ED (d 30-31). Volunteers lost (P<0.05) 3.2 ± 0.2 kg body weight during ED regardless of dietary protein. The proportion of weight loss due to reductions in fat-free mass was lower (P<0.05) and the loss of fat mass was higher (P<0.05) in those receiving 2×-RDA and 3×-RDA compared to RDA. The anabolic muscle response to a protein-rich meal during ED was not different (P>0.05) from WM for 2×-RDA and 3×-RDA, but was lower during ED than WM for those consuming RDA levels of protein (energy × protein interaction, P<0.05). To assess muscle protein metabolic responses to varied protein intakes during ED, RDA served as the study control. In summary, we determined that consuming dietary protein at levels exceeding the RDA may protect fat-free mass during short-term weight loss.”

[22] Evidence type: randomised prospective study

Flechtner-Mors M, Boehm BO, Wittmann R, Thoma U, Ditschuneit HH.
Diabetes Metab Res Rev. 2010 Jul;26(5):393-405. doi: 10.1002/dmrr.1097.

“Obese subjects received instructions for an energy-restricted diet with a calorie deficit of 500 kcal/day and were randomly assigned to either high-protein (1.34 g/kg body weight) or conventional protein (0.8 g/kg body weight) diets for 12 months. Protein-enriched meal replacements were used to enrich one arm of the diet with protein throughout the study. In all, 67% of the participants completed the 1-year study.
Subjects following the high-protein diet lost more body weight and more fat mass compared with those on the conventional protein diet, whereas the loss of fat-free mass was similar in both diet groups. Biochemical parameters associated with the metabolic syndrome improved in both diet groups. Improvements were modestly greater in subjects with the high-protein diet. After 12 months of treatment, 64.5% of the subjects in the high-protein diet group and 34.8% of the subjects in the conventional diet group no longer met three or more of the criteria for having the metabolic syndrome.”

[23] Evidence type: explanation and comparison of experiments.

Stephen D Phinney.
Nutrition & Metabolism 2004, 1:2 doi:10.1186/1743-7075-1-2

“An example of what happens when these mineral considerations are not heeded can be found in a study prominently published in 1980 [18]. This was a study designed to evaluate the relative value of “protein only” versus “protein plus carbohydrate” in the preservation of lean tissue during a weight loss diet. The protein only diet consisted solely of boiled turkey (taken without the broth), whereas the protein plus carbohydrate consisted of an equal number of calories provided as turkey plus grape juice. Monitored for 4 weeks in a metabolic ward, the subjects taking the protein plus carbohydrate did fairly well at maintaining lean body mass (measured by nitrogen balance), whereas those taking the protein only experienced a progressive loss of body nitrogen.
“A clue to what was happening in this “Turkey Study” could be found in the potassium balance data provided in this report. Normally, nitrogen and potassium gains or losses are closely correlated, as they both are contained in lean tissue. Interestingly, the authors noted that the protein only diet subjects were losing nitrogen but gaining potassium. As noted in a rebuttal letter published soon after this report [19], this anomaly occurred because the authors assumed the potassium intake of their subjects based upon handbook values for raw turkey, not recognizing that half of this potassium was being discarded in the unconsumed broth. Deprived of this potassium (and also limited in their salt intake), these subjects were unable to benefit from the dietary protein provided and lost lean tissue. Also worthy of note, although this study was effectively refuted by a well-designed metabolic ward study published 3 years later [20], this “Turkey Study” continues to be quoted as an example of the limitations of low carbohydrate weight loss diets.”

[24] Evidence Type: experiment

Volek, Jeff S., Erin E. Quann, and Cassandra E. Forsythe.
Strength & Conditioning Journal 32.1 (2010): 42-47.

“We performed a similar experiment in overweight/obese men who were placed in a low-fat diet group that restricted fat to less than 25% of energy or a very low–carbohydrate ketogenic diet group that reduced carbohydrate to less than 15% energy. Both groups also participated in a resistance training program (see Practical Applications) (16). Body composition was assessed using dual energy x-ray absorptiometry before and after the 12-week program. The results were compared with non-training diet only groups. As expected, the low-carbohydrate diet group lost more fat, which was associated with greater decreases in insulin. Resistance training, independent of diet, resulted in increased lean body mass without compromising fat loss in both diet groups. The most dramatic reduction in percent body fat was in the low-carbohydrate diet resistance training group (25.3%), followed by low-fat resistance training(23.5%), low-carbohydrate diet only (23.4%), and low-fat diet only (22.0%) groups. These data show for the first time that resistance training is a potent stimulus to protect lean body mass in men consuming a low-carbohydrate diet, while still allowing for significantly greater fat loss.”

[25] Evidence type: summary of experiments

PG Lindner, GL Blackburn.
Obesity/Bariatric Med, 1976, pp. 198–216

“The question that remained to be answered was what amount of protein was necessary to maintain positive energy nitrogen balance ⁹³ for a time sufficient to allow the obese individual to reach his normal size.
“Blackburn, et al investigated this question while also measuring the biochemical and clinical aspects of what he termed the protein-sparing modified fast (PSMF). One hundred forty-six patients (111 outpatients) were followed at the Massachusetts Institute of Technology Clinical Research Center while undergoing a comprehensive weight reduction program that included nutritional education, behavior modification, exercise, and the PSMF. ⁹⁴’⁹⁵ He found that in males, nitrogen (N) balance was obtained at 1.4 gm protein daily/kg ideal body weight (IBW) and in females, N balance was obtained at 1.2 gm/kg IBW. Outpatients tolerated the PSMF well in a wide variety of lifestyles and occupations and invariably found it easier than a balanced deficit diet. Ingestion of 40 to 80 gm of carbohydrate in addition to protein, however, markedly reduced ketosis and reistated hunger.⁹⁶”

similarities between germ-free mice and ketogenic humans

similarities between germ-free mice and ketogenic humans

tracing a chain of ideas

Sometimes the assumptions that scientists start with about what is “good”, “healthy”, or “normal”
can cause them to interpret results in a completely different way than someone starting with different assumptions would have.
Then, the resulting conclusions become the assumptions in the next round of interpretation,
leading to a chain of logic in which one questionable assumption leads to another.
We recently read a paper in which the authors made a series of logical steps,
and it became almost comical to us how at each step we would have interpreted
their results in an opposite way than they did.
When the results of their experiment are looked at from our perspective, it
suggests an intriguing hypothesis:
Maybe some of the health benefits a ketogenic diet are due, not just to the
diet being low in digestible carbohydrate and thus leading to ketosis, but
also to being low in indigestible fiber and thus starving certain gut

Or, to phrase the same hypothesis differently, maybe one mechanism by which a
glycolytic or high-fiber diet causes health problems is that it feeds harmful gut bacteria,
and the presence of those bacteria causes the health problems.
If that hypothesis were true, it would imply that if you are eating a
low-carb diet, then including a lot of low-carb vegetables would feed these
hypothesized harmful gut bacteria and reduce some of the potential health benefits of
a low-carb diet.

in brief

The purpose of this article is two-fold:

  • First, to compare the authors’ interpretations of the observations to ours, given what we know about the metabolic effects of ketogenic
    We draw attention to the fact that the metabolism of germ-free mice is strikingly similar to that of ketogenic dieters.
    This similarity holds at the whole-body level in terms of behaviour and physical characteristics, as well as the level of mitochondrial
    We show that these characteristics appear to be beneficial.
  • Second, to raise the following questions:
    Are some of the benefits of a ketogenic diet mediated by starving gut bacteria, and if so,
    does eating fiber (i.e. low-carb vegetables) reduce some of the health benefits of a keto diet?
    Would eating a carbohydrate- and fiber- free diet confer some keto-like benefits even in the absence of ketosis?

the end of the chain

It all started when we saw the following statement on the Wikipedia page about butyrate :

"Butyrates are important as food for cells lining the mammalian colon
(colonocytes). Without butyrates for energy, colon cells undergo
autophagy (self digestion) and die.[1] Short-chain fatty acids, which
include butyrate, are produced by beneficial colonic bacteria
(probiotics) that feed on, or ferment prebiotics, which are plant
products that contain adequate amounts of dietary fiber."

The topic of butyrate is exciting to some scientists, because they have the idea that eating indigestible fiber is good for human health.
Epidemiological studies have found correlations between high fiber intake and relatively less disease.
However, randomised controlled trials have repeatedly failed to confirm the hypotheses that the fiber intake was actually protective. [1].
Therefore when a new idea comes up that might explain how eating fiber would be good for human health, scientists still hoping for such evidence latch onto it.
Butyrate is an example of such a candidate mechanism for how eating fiber would be good for human health.
To us, since we think that eating fiber is useless (at best) for health,
the statement above poses a challenge and a mystery.
Almost the only source of butyrate in the human body is, as the wikipedia page explains in the excerpt above,
from gut bacteria digesting fiber that you ate.
(There’s also some butyrate in butter, presumably made the same way in cows.)
If butyrate is necessary for the health and even the survival of colon cells,
wouldn’t that mean that a low-fiber diet — such as an all-meat diet — would
be very unhealthy?
Amber hasn’t eaten any significant amount of indigestible fiber in more than four years;
does this mean that her colon cells have died off?
So we set out to investigate what led to that statement on wikipedia.
Our investigation ultimately led to an intriguing hypothesis about a candidate mechanism to explain some of the health benefits of a keto diet.
The paper referenced as "[1]" on the wikipedia page is Donohoe-2011-“The Microbiome and Butyrate Regulate Energy Metabolism and Autophagy in the Mammalian Colon”. We read it with interest.
The authors of this paper performed a good experiment, made precise measurements, got interesting results, and clearly reported their results.
But when it came to interpretation, they started with some assumptions we don’t think are warranted,
and therefore produced a chain of reasoning that eventually led them and their readers, such as the authors of the wikipedia page, to conclusions that are opposite from ours.

similarities between germ-free mice and ketogenic dieters

If you have followed the debates about low-carb diets conferring a metabolic advantage (demonstrated by their superior performance as weight loss diets [2], [3])
then the above description of germ-free mice should sound familiar.
Compared to low-fat dieters, ketogenic dieters tend to be leaner (i.e. have a higher ratio of muscle to body fat), and have lower insulin and blood glucose levels.
This can happen despite similar caloric intake.
Their liver glycogen levels are also lower; ketogenesis may depend on low glycogen levels [4].
However, the germ-free mice are not ketogenic, presumably because they are consuming regular, glucose-plentiful diets.
In fact they are less ketogenic than the conventional mice, as measured by beta-hydroxybutyrate in the blood.
So what is the cause of the similarity in metabolism between germ-free mice and ketogenic humans?

There is one other important way in which the germ-free mice were different from the conventional mice.
They had lower NADH/NAD+ ratios and ATP levels (per mitochondrion) in their colon cells, but not in the liver, heart, or kidneys.
Note that the heart, liver and kidneys favour fat metabolism, even in glycolytic (non-ketogenic) dieters [5].
The authors took this to be further evidence that the mice were in a state of energy deprivation,
even though by all accounts they appeared to be using substantially more energy.
They even previously mentioned this in connection with the reduced fatness:

"[Germ-free] mice exhibit increased locomotor activity. Therefore, the increased food consumption and decreased body fat of germ-free mice may simply be due to increased energy expenditure." — from Donohoe-2011

However, it is a mistake to assume that lower NADH/NAD+ ratios and ATP levels per mitochondrion corresponds to less cellular energy.
In fact, it is likely to be the opposite.

mitochondrial energetics is the commonality

There are three other conditions we know of that reduce the NADH/NAD+ ratio: calorie restriction [6], ketogenic diets [7], and the diabetes drug metformin [8].
Ketogenic diets share mechanisms with caloric restriction.
Indeed, it seems likely that benefits of calorie restriction come from the activation of ketone bodies [9].
When you use fat and ketones for fuel instead of glucose, you produce fewer free radicals through reducing the NADH/NAD+ ratio [6].
This is probably the main mechanism by which it achieves the neuroprotection we mentioned in a recent post.
Similarly, the reduction of the NADH/NAD+ ratio is probably one of the mechanisms by which calorie restriction can increase lifespan [5].
Calorie restriction also preserves ATP production, but it does this by increasing the number of mitochondria to match or exceed the lower ATP yield per mitochondrion [10], [11].
There is preliminary evidence that a ketogenic diet also increases mitochondrial number [12], [13].
So the idea of Donohoe et al. — that total ATP production is compromised because the NADH/NAD+ ratio in the individual mitochondrion has lowered ATP output per mitochondrion — seems unwarranted.
Instead, there is likely to be a compensatory increase in mitochondrial number.
That would be consistent with the fact that the mice appear to have more energy, not less.
Ketogenic dieters have also been measured to have more energy expenditure than low-fat dieters [14].
So, a reduction of NADH/NAD+ ratio is associated with health benefits, and proposed longevity mechanisms.
As you might now suspect, previous studies have shown that germ-free animals have increased lifespans [15].
(They also show decreased anxiety [16] and increased bone mass [17]. Once again, to us this sounds like a better kind of mouse to be!)

fiber-free for better health?

Given this observation — that some of the benefits of ketogenic diets are present in mice that don’t have gut bacteria which process dietary fiber, even though the mice are not in ketosis — it raises the following questions:

  1. Could the starvation of gut bacteria be a part of the mechanism of the benefits of ketogenic diets?
  2. Since butyrate restores the mitochondrial working of the cells to be like the conventional controls (which, from our perspective, is a worse physiological state), could fiber be actually counter-productive to a ketogenic diet?
  3. In analogy to the way the putative benefits of fiber may simply be that they displace refined carbohydrates in the diet, could the reason probiotics can lead to improved health be not because they are beneficial, but because they push out more harmful strains [18]?
  4. Could a diet free of carb and fiber (i.e. one extremely low in plants) have benefits independent of its tendency to be ketogenic?

We don’t have enough evidence to settle these questions, but they are interesting hypotheses that come directly from the results of this study.

in sum

  • Contrary to the conclusions of the authors and Wikipedia editors that butyrate is necessary for cell energy, we interpret the results as showing improved cellular energy in the absence of butyrate.
  • We now have another source for making hypotheses about potentially important cellular metabolism. Before we learned about germ-free mice, we could already use mechanisms discovered from caloric restriction and compare them to mechanisms of ketogenic diets. Now we can compare and contrast mechanisms from caloric restriction, ketogenic diets, and germ-free animals.
  • This raises some interesting (perhaps even provocative) questions about the health effects of dietary fiber and gut flora on human health.


[1] Evidence type: review

Carla S Coffin, MD FRCPC and Eldon A Shaffer, MD FRCPC
Can J Gastroenterol. 2006 April; 20(4): 255–256.

(emphasis ours)
"A recent pooled analysis of 13 prospective cohort studies (6) found that dietary fibre was not associated with a reduced risk of colorectal cancer after adjusting for other dietary risk factors. The Cochrane collaboration (7) systematically reviewed five studies of over 4000 subjects for the effect of dietary fibre on the incidence or recurrence of colorectal adenomas and incidence of colorectal cancer over a two-to four-year period. The population included all subjects that had adenomatous polyps but no history of colorectal cancer or a documented ‘clean colon’ at baseline with follow-up colonoscopy. Study interventions included soluble and insoluble dietary fibre or a comprehensive dietary intervention with high fibre whole food sources. The combined data showed no outcome difference between the intervention and control groups in the number of subjects with at least one adenoma or a new diagnosis of colorectal cancer. The Cochrane reviewers (7) concluded that there was no evidence from randomized controlled trials to suggest that increased dietary fibre intake would reduce the incidence or recurrence of adenomatous polyps.
"Widespread popular media advertisements have purported the benefits of soluble fibre in lowering the risk of atherosclerotic coronary artery disease, mainly by modifying the main coronary artery disease risk factors (ie, dyslipidemia, diabetes and obesity). As for diabetes, high fibre diets slow the postprandial rise in blood glucose and thus, improve glycemic control (8). In dyslipidemic patients, pundits have proposed that psyllium lowers serum cholesterol by binding bile acids in the intestinal lumen resulting in decreased absorption and increased fecal excretion. The ensuing bile acid depletion increases hepatic demand for the de novo synthesis of bile acids from cholesterol. Investigating this mechanism, Van Rosendaal et al (9) found that fibre administration had no effect and certainly did not lower serum cholesterol. Similarly, an earlier study (10) comparing the effect of wheat bran on serum cholesterol of hyperlipidemic and normolipidemic controls showed no change in total cholesterol or ratio of low density lipoprotein to high density lipoprotein cholesterol. Another trial (11) of intensive dietary advice regarding fat, cereal fibre and fish intake on diet and mortality of men with a recent history of myocardial infarction did not find any substantial long-term benefit. The authors admitted to limitations of dietary data in the study (ie, only short-term period of advice and limited number of questions), but there was no evidence to guide decisions about value of dietary advice to increase fish or cereal fibre by people with coronary disease. We await the results of three Cochrane protocols undertaken to review the evidence of dietary fibre in fruits and vegetables, wholegrain cereals or high-fat, low fibre dietary intervention in the prevention of coronary heart disease (12–14). Any conclusions regarding the effectiveness of fibre for the prevention of heart disease appear premature."
"In one of the first randomized, placebo-controlled trials of the role of bran in patients with [diverticular disease] (17), the authors concluded that dietary fibre supplements do nothing more than relieve constipation, and the impression that fibre helps [diverticular disease] is “simply a manifestation of western civilization’s obsession with the need for frequent defecation”. Recent systematic reviews (18,19) of the role of dietary fibre and [diverticular disease] (both asymptomatic diverticulosis and symptomatic diverticulitis) conclude that most of the positive evidence of the effects of fibre supplementation in treating or preventing disease is from retrospective analyses with inherent limitations and high risk of bias."
"Systematic reviews have shown that the treatment of IBS patients with fibre is controversial. One recent meta-analysis of 17 randomized controlled trials (20) quantified the effectiveness of different types of fibre. The reviewers found that fibre was only marginally effective in terms of global symptom improvement or constipation and there was no effect in IBS related abdominal pain. Fibre has a role in treating constipation but its value for IBS, pain and diarrhea is controversial. Any effectivenss of fibre in the long-term management of IBS remains questionable. Clinically, bran is no better than placebo in the relief of the overall symptoms of IBS, and is possibly worse than a normal diet for some symptoms."

[2] Evidence type: review of randomised controlled trials in humans.

Hession M, Rolland C, Kulkarni U, Wise A, Broom J.
Obes Rev. 2009 Jan;10(1):36-50. doi: 10.1111/j.1467-789X.2008.00518.x. Epub 2008 Aug 11.

There are few studies comparing the effects of low-carbohydrate/high-protein diets with low-fat/high-carbohydrate diets for obesity and cardiovascular disease risk. This systematic review focuses on randomized controlled trials of low-carbohydrate diets compared with low-fat/low-calorie diets. Studies conducted in adult populations with mean or median body mass index of > or =28 kg m(-2) were included. Thirteen electronic databases were searched and randomized controlled trials from January 2000 to March 2007 were evaluated. Trials were included if they lasted at least 6 months and assessed the weight-loss effects of low-carbohydrate diets against low-fat/low-calorie diets. For each study, data were abstracted and checked by two researchers prior to electronic data entry. The computer program Review Manager 4.2.2 was used for the data analysis. Thirteen articles met the inclusion criteria. There were significant differences between the groups for weight, high-density lipoprotein cholesterol, triacylglycerols and systolic blood pressure, favouring the low-carbohydrate diet. There was a higher attrition rate in the low-fat compared with the low-carbohydrate groups suggesting a patient preference for a low-carbohydrate/high-protein approach as opposed to the Public Health preference of a low-fat/high-carbohydrate diet. Evidence from this systematic review demonstrates that low-carbohydrate/high-protein diets are more effective at 6 months and are as effective, if not more, as low-fat diets in reducing weight and cardiovascular disease risk up to 1 year. More evidence and longer-term studies are needed to assess the long-term cardiovascular benefits from the weight loss achieved using these diets."

[3] Evidence type: review of randomised controlled trials in humans.

Although this analysis is not peer-reveiwed, it is thorough, appears to be accurate, and does not omit any counter-evidence as far as we are aware.
Kris Gunnars
October 15, 2013
Authority Nutrition

(emphasis ours)
"In this article, I have analyzed the data from 23 of these studies comparing low-carb and low-fat diets.
"All of the studies are randomized controlled trials, the gold standard of science. All are published in respected, peer-reviewed journals.

"The majority of studies achieved statistically significant differences in weight loss (always in favor of low-carb). There are several other factors that are worth noting:

  • The low-carb groups often lost 2-3 times as much weight as the low-fat groups. In a few instances there was no significant difference.
  • In most cases, calories were restricted in the low-fat groups, while the low-carb groups could eat as much as they wanted.
  • When both groups restricted calories, the low-carb dieters still lost more weight (7, 13, 19), although it was not always significant (8, 18, 20).
  • There was only one study where the low-fat group lost more weight (23) although the difference was small (0.5 kg – 1.1 lb) and not statistically significant.
  • In several of the studies, weight loss was greatest in the beginning. Then people start regaining the weight over time as they abandon the diet.
  • When the researchers looked at abdominal fat (the unhealthy visceral fat) directly, low-carb diets had a clear advantage (5, 7, 19).
[4] Evidence type: review of experiments

McGarry JD, Foster DW.
Arch Intern Med. 1977 Apr;137(4):495-501.

A two-site, bihormonal concept for the control of ketone body production is proposed. Thus, ketosis is viewed as the result of increased mobilization of free fatty acids from adipose tissue (site 1) to the liver (site 2), coupled with simultaneous enhancement of the liver’s capacity to convert these substrates into acetoacetic and beta-hydroxybutyric acids. The former event is believed to be triggered by a fall in plasma insulin levels while the latter is considered to be effected primarily by the concomitant glucagon excess characteristic of the ketotic state. Although the precise mechanism whereby elevation of the circulating [glucagon]:[insulin] ratio stimulates hepatic ketogenic potential is not known, activation of the carnitine acyltransferase reaction, the first step in the oxidation of fatty acids, is an essential feature. Two prerequisites for this metabolic adaptation in liver appear to be an elevation in its carnitine content and depletion of its glycogen stores. Despite present limitations the model (evolved mainly from rat studies) provides a framework for the description of various types of clinical ketosis in biochemical terms and may be useful for future studies."

[5] Evidence type: authority

El Bacha, T., Luz, M. & Da Poian, A. (2010)
Nature Education 3(9):8

"[M]any different cells do oxidize fatty acids for ATP production. Between meals, cardiac muscle cells meet 90% of their ATP demands by oxidizing fatty acids. Although these proportions may fall to about 60% depending on the nutritional status and the intensity of contractions, fatty acids may be considered the major fuel consumed by cardiac muscle.

Other organs that use primarily fatty acid oxidation are the kidney and the liver."

[6] Evidence type: controlled non-human animal experiments

Su-Ju Lin, Ethan Ford, Marcia Haigis, Greg Liszt, and Leonard Guarente.
Genes Dev. 2004 January 1; 18(1): 12–16.

"Our studies show that a switch to oxidative metabolism during CR increases the NAD/NADH ratio by decreasing NADH levels. NADH is a competitive inhibitor of Sir2, implying that a reduction in this dinucleotide activates Sir2 to extend the life span in CR. Indeed, overexpression of the NADH dehydrogenase specifically lowers NADH levels and extends the life span, providing strong support for this hypothesis. Regulation of the life span by NADH is also consistent with the earlier finding that electron transport is required for longevity during CR (Lin et al. 2002). The NAD/NADH ratio reflects the intracellular redox state and is a readout of metabolic activity. Our findings suggest that this ratio can serve a critical regulatory function, namely, the determination of the life span of yeast mother cells. It remains to be seen whether this ratio will serve related regulatory functions in higher organisms."

[7] Evidence type: controlled non-human animal experiments

Marwan Maalouf, Patrick G. Sullivan, Laurie Davis, Do Young Kim, and Jong M. Rho Neuroscience.
2007 March 2; 145(1): 256–264.

"[W]e demonstrate that ketones reduce glutamate-induced free radical formation by increasing the NAD+/NADH ratio and enhancing mitochondrial respiration in neocortical neurons. This mechanism may, in part, contribute to the neuroprotective activity of ketones by restoring normal bioenergetic function in the face of oxidative stress."

[8] Evidence type: controlled non-human animal experiment

Paul W Caton, Nanda K Nayuni, Julius Kieswich, Noorafza Q Khan, Muhammed M Yaqoob and Roger Corder
J Endocrinol April 1, 2010 205 97-106

(emphasis ours)
Metformin increases SIRT1 in db/db mice
Systemic activation of SIRT1 with the activator SRT1720 is reported to lower blood glucose and improve insulin sensitivity in Zucker rats and diet-induced obese mice in part through inhibition of hepatic gluconeogenesis (Milne et al. 2007). Therefore, we investigated whether metformin inhibited gluconeogenesis through changes in hepatic SIRT1. Eight-week-old db/db or control (db/m) mice were administered metformin (250 mg/kg per day; 7 days). Levels of SIRT1 protein, activity and NAD+/NADH ratio were significantly increased in metformin-treated db/db mice compared with the controls and untreated db/db mice (Fig. 1A, C and D). Despite increased protein levels, Sirt1 mRNA levels were unchanged following metformin treatment (Fig. 1B). Levels of SIRT1 protein and activity as well as NAD+/NADH levels were unchanged between the control and untreated mice (Fig. 1A–C). Metformin had no effect on SIRT1 in control mice (data not shown). Furthermore, incubation of HepG2 cells with metformin (2 mM) also resulted in increased levels of SIRT1 protein and activity and NAD+/NADH ratio (Fig. 1E–G). This indicates that increasing SIRT1 protein and activity could be a key mechanism by which metformin inhibits gluconeogenic gene expression."

[9] Evidence type: review of experiments

Marwan A. Maalouf, Jong M. Rho, and Mark P. Mattson
Brain Res Rev. 2009 March; 59(2): 293–315.

"Calorie restriction and the ketogenic diet share two characteristics: reduced carbohydrate intake and a compensatory rise in ketone bodies. The neuroprotective effects of reduced carbohydrate per se are being investigated by several research groups (Mattson et al. 2003; Ingram et al. 2006). We have evaluated the possibility that ketone bodies might mediate the neuroprotective effects of calorie restriction and of the ketogenic diet. An expanding body of evidence indicates that ketone bodies are indeed neuroprotective and that the underlying mechanisms are similar to those associated with calorie restriction – specifically at the mitochondrial level."

[10] Evidence type: review of clinical reports

G. López-Lluch, N. Hunt, B. Jones, M. Zhu, H. Jamieson, S. Hilmer, M. V. Cascajo, J. Allard, D. K. Ingram, P. Navas, and R. de Cabo
Proc Natl Acad Sci U S A. 2006 February 7; 103(6): 1768–1773.

"[M]itochondria under CR conditions show less oxygen consumption, reduce membrane potential, and generate less reactive oxygen species than controls, but remarkably they are able to maintain their critical ATP production. In effect, CR can induce a peroxisome proliferation-activated receptor coactivator 1α-dependent increase in mitochondria capable of efficient and balanced bioenergetics to reduce oxidative stress and attenuate age-dependent endogenous oxidative damage."

[11] Evidence type: review of controlled experiments

Marwan A. Maalouf, Jong M. Rho, and Mark P. Mattson
Brain Res Rev. 2009 March; 59(2): 293–315.

(emphasis ours)
"Slowing of brain aging in calorie-restricted animals was originally believed to result from reduced metabolic activity and, hence, decreased production of reactive oxygen species, a natural byproduct of oxidative metabolism (Wolf 2006). Several studies revealed that calorie restriction was associated with energy conservation (Gonzales-Pacheco et al. 1993; Santos-Pintos et al. 2001) and that mitochondria isolated from calorie-restricted animals produced less ATP than those from controls fed ad libitum, a finding compatible with increased UCP activity (Sreekumar et al. 2002; Drew et al. 2003). However, separate investigations in rodents have suggested that, when adjusted for body weight, metabolic rate does not decrease with calorie restriction (Masoro et al. 1982; McCarter et al. 1985; Masoro 1993). More importantly, calorie restriction prevents the age-related decline in oxidative metabolism in muscle (Hepple et al. 2005; Baker et al. 2006). These data are supported by recent studies indicating that, in contrast to isolated mitochondria, ATP synthesis in intact myocytes and in vivo does not decrease following calorie restriction (Lopez-Lluch et al. 2006; Zangarelli et al. 2006). Additional support is provided by the finding that, in yeast, oxidative metabolism increases with calorie restriction (Lin et al. 2002). […] Although the effects of calorie restriction on ATP generation might appear to contradict those on uncoupling proteins, this discrepancy can be explained by the fact that calorie restriction also promotes mitochondrial biogenesis, thereby enhancing total metabolic output per cell while decreasing mitochondrial production of reactive oxygen species (Diano et al. 2003; Nisoli et al. 2005; Civitarese et al. 2007)."

[12] Evidence type: in vitro non-human animal experiment

Bough KJ, Wetherington J, Hassel B, Pare JF, Gawryluk JW, Greene JG, Shaw R, Smith Y, Geiger JD, Dingledine RJ.
Ann Neurol. 2006 Aug;60(2):223-35.

The full anticonvulsant effect of the ketogenic diet (KD) can require weeks to develop in rats, suggesting that altered gene expression is involved. The KD typically is used in pediatric epilepsies, but is effective also in adolescents and adults. Our goal was to use microarray and complementary technologies in adolescent rats to understand its anticonvulsant effect.
Microarrays were used to define patterns of gene expression in the hippocampus of rats fed a KD or control diet for 3 weeks. Hippocampi from control- and KD-fed rats were also compared for the number of mitochondrial profiles in electron micrographs, the levels of selected energy metabolites and enzyme activities, and the effect of low glucose on synaptic transmission.
Most striking was a coordinated upregulation of all (n = 34) differentially regulated transcripts encoding energy metabolism enzymes and 39 of 42 transcripts encoding mitochondrial proteins, which was accompanied by an increased number of mitochondrial profiles, a higher phosphocreatine/creatine ratio, elevated glutamate levels, and decreased glycogen levels. Consistent with increased energy reserves, synaptic transmission in hippocampal slices from KD-fed animals was resistant to low glucose.
These data show that a calorie-restricted KD enhances brain metabolism. We propose an anticonvulsant mechanism of the KD involving mitochondrial biogenesis leading to enhanced alternative energy stores."

[13] Evidence type: controlled non-human animal experiments

Srivastava S, Kashiwaya Y, King MT, Baxa U, Tam J, Niu G, Chen X, Clarke K, Veech RL.
FASEB J. 2012 June; 26(6): 2351–2362.

(Emphasis ours)
"We measured the effects of a diet in which d-β-hydroxybutyrate-(R)-1,3 butanediol monoester [ketone ester (KE)] replaced equicaloric amounts of carbohydrate on 8-wk-old male C57BL/6J mice. Diets contained equal amounts of fat, protein, and micronutrients. The KE group was fed ad libitum, whereas the control (Ctrl) mice were pair-fed to the KE group. Blood d-β-hydroxybutyrate levels in the KE group were 3-5 times those reported with high-fat ketogenic diets. Voluntary food intake was reduced dose dependently with the KE diet. Feeding the KE diet for up to 1 mo increased the number of mitochondria and doubled the electron transport chain proteins, uncoupling protein 1, and mitochondrial biogenesis-regulating proteins in the interscapular brown adipose tissue (IBAT). [18F]-Fluorodeoxyglucose uptake in IBAT of the KE group was twice that in IBAT of the Ctrl group. Plasma leptin levels of the KE group were more than 2-fold those of the Ctrl group and were associated with increased sympathetic nervous system activity to IBAT. The KE group exhibited 14% greater resting energy expenditure, but the total energy expenditure measured over a 24-h period or body weights was not different. The quantitative insulin-sensitivity check index was 73% higher in the KE group. These results identify KE as a potential antiobesity supplement."

[14] Evidence type: randomised controlled clinical trial

Cara B. Ebbeling, PhD; Janis F. Swain, MS, RD; Henry A. Feldman, PhD; William W. Wong, PhD; David L. Hachey, PhD; Erica Garcia-Lago, BA; David S. Ludwig, MD, PhD
JAMA. 2012;307(24):2627-2634. doi:10.1001/jama.2012.6607.

"The results of our study challenge the notion that a calorie is a calorie from a metabolic perspective. During isocaloric feeding following weight loss, REE was 67 kcal/d higher with the very low-carbohydrate diet compared with the low-fat diet. TEE differed by approximately 300 kcal/d between these 2 diets, an effect corresponding with the amount of energy typically expended in 1 hour of moderate-intensity physical activity."

[15] Evidence type: review of non-human animal experiments

H A Gordon and L Pesti.
Bacteriol Rev. 1971 December; 35(4): 390–429.

"Two attempts have been made to construct life tables and to determine lesions at natural death in germ-free and conventional animals. One study (105) was conducted in genetically closely linked Swiss Webster mice and included over 300 germ-free and the same number of conventional controls which were introduced into the colony at the age of 12 months (to eliminate the effect of early losses). At natural death, the ages of the mice were (means and standard errors in days 19; females, are given): germ-free males, 723 681 i 12; conventional males, 480 i 10; females, 516 i 10. This pattern of survival rates seemed to continue throughout the course of the experi- ment. In the second study (335), approximately 50 germ-free and the same number of conven- tional ICR mice were introduced into the colony after weaning. At natural death, the age of the mice was (using the same mode of expression): germ-free males, 556 i 43; females, 535 + 46; 41; females, 547 ± conventional males, 536 45. In the first trimester of life, the survival rate was essentially the same in the germ-free and conventional control groups. In the middle third, the germ-free mice displayed an increased survival rate (e.g., 40% cumulative mortality was reached for the combined group of germ-free males and females only at the age of approximately 580 days, whereas for the conventional controls this value was approximately 410 days). Increased mortality of the germ-free group at more advanced age resulted in the similarity of mean ages between the opposing animal groups when all animals participating in the study were considered."

[16] Evidence type: controlled non-human animal experiment

Rochellys Diaz Heijtz, Shugui Wang, Farhana Anuar, Yu Qian, Britta Björkholm, Annika Samuelsson, Martin L. Hibberd, Hans Forssberg, and Sven Pettersson
Proc Natl Acad Sci U S A. 2011 February 15; 108(7): 3047–3052.

"Microbial colonization of mammals is an evolution-driven process that modulate host physiology, many of which are associated with immunity and nutrient intake. Here, we report that colonization by gut microbiota impacts mammalian brain development and subsequent adult behavior. Using measures of motor activity and anxiety-like behavior, we demonstrate that germ free (GF) mice display increased motor activity and reduced anxiety, compared with specific pathogen free (SPF) mice with a normal gut microbiota. This behavioral phenotype is associated with altered expression of genes known to be involved in second messenger pathways and synaptic long-term potentiation in brain regions implicated in motor control and anxiety-like behavior. GF mice exposed to gut microbiota early in life display similar characteristics as SPF mice, including reduced expression of PSD-95 and synaptophysin in the striatum. Hence, our results suggest that the microbial colonization process initiates signaling mechanisms that affect neuronal circuits involved in motor control and anxiety behavior."

[17] Evidence type: controlled non-human animal experiment

Sjögren K, Engdahl C, Henning P, Lerner UH, Tremaroli V, Lagerquist MK, Bäckhed F, Ohlsson C.
J Bone Miner Res. 2012 Jun;27(6):1357-67. doi: 10.1002/jbmr.1588.

The gut microbiota modulates host metabolism and development of immune status. Here we show that the gut microbiota is also a major regulator of bone mass in mice. Germ-free (GF) mice exhibit increased bone mass associated with reduced number of osteoclasts per bone surface compared with conventionally raised (CONV-R) mice. Colonization of GF mice with a normal gut microbiota normalizes bone mass. Furthermore, GF mice have decreased frequency of CD4(+) T cells and CD11b(+) /GR 1 osteoclast precursor cells in bone marrow, which could be normalized by colonization. GF mice exhibited reduced expression of inflammatory cytokines in bone and bone marrow compared with CONV-R mice. In summary, the gut microbiota regulates bone mass in mice, and we provide evidence for a mechanism involving altered immune status in bone and thereby affected osteoclast-mediated bone resorption. Further studies are required to evaluate the gut microbiota as a novel therapeutic target for osteoporosis."

[18] Evidence type: review of experiments and hypotheses

Parvez S, Malik KA, Ah Kang S, Kim HY.
J Appl Microbiol. 2006 Jun;100(6):1171-85.

(emphasis ours)
Probiotics are usually defined as microbial food supplements with beneficial effects on the consumers. Most probiotics fall into the group of organisms’ known as lactic acid-producing bacteria and are normally consumed in the form of yogurt, fermented milks or other fermented foods. Some of the beneficial effect of lactic acid bacteria consumption include: (i) improving intestinal tract health; (ii) enhancing the immune system, synthesizing and enhancing the bioavailability of nutrients; (iii) reducing symptoms of lactose intolerance, decreasing the prevalence of allergy in susceptible individuals; and (iv) reducing risk of certain cancers. The mechanisms by which probiotics exert their effects are largely unknown, but may involve modifying gut pH, antagonizing pathogens through production of antimicrobial compounds, competing for pathogen binding and receptor sites as well as for available nutrients and growth factors, stimulating immunomodulatory cells, and producing lactase. Selection criteria, efficacy, food and supplement sources and safety issues around probiotics are reviewed. Recent scientific investigation has supported the important role of probiotics as a part of a healthy diet for human as well as for animals and may be an avenue to provide a safe, cost effective, and ‘natural’ approach that adds a barrier against microbial infection. This paper presents a review of probiotics in health maintenance and disease prevention."

The Ketogenic Diet Reverses Indicators of Heart Disease

sup {font-size: .6em;}

The Ketogenic Diet Reverses Indicators of Heart Disease

Cardiovascular disease (CVD) is the leading cause of death worldwide
Because of its prevalence and life-threatening nature, and because it appears that a keto diet is likely to reverse it, we consider it one of the most important conditions to discuss here.

In our last post, we argued that CVD, being a disease strongly associated with metabolic syndrome, is likely to be best treated with a ketogenic diet.
In this post we will present more evidence that ketogenic diets do improve heart disease risk factors.

Unfortunately, there is much confusion and misinformation about the impact of nutrition on CVD among scientists and non-scientists alike.
Not only does a high fat, keto diet not worsen heart disease risk — as would commonly be assumed — it actually improves it.
This confusion about dietary fat is probably the reason that we do not yet have clinical trials directly testing the effects of ketogenic diets on CVD outcomes.

However, we already have many trials of ketogenic diets that measured known CVD risk factors, especially cholesterol profiles.
It turns out that these trials show a powerful heart disease risk reduction in those following a ketogenic diet.
It is powerful both in absolute terms, and in comparison with low-fat diets, which tend to improve some weakly predictive factors while worsening stronger predictors.

As such, a high-fat ketogenic diet is currently the best known non-drug intervention for heart disease, as defined by mainstream measures of risk. It is arguably better than drug interventions, too.

In brief:

  • Total cholesterol and LDL cholesterol are only weak predictors of CVD.
  • Triglycerides, HDL, LDL particle size, and the HDL-to-triglyceride ratio are much stronger predictors of CVD.
  • Keto diets improve triglyceride levels, HDL, and LDL particle size — precisely those measures that strongly indicate risk.

Total cholesterol and LDL cholesterol are only weakly associated with CVD

The connection between blood cholesterol levels and the development of heart disease began to be explored in the last century.
Over the last several decades, our understanding of the predictive power of various blood lipids has gone through many refinements as our ability to measure finer and finer detail has advanced.

In the early years, it appeared that high levels of total cholesterol carried some risk of heart disease in many cases.
However, it is now well established that total cholesterol by itself is a weak predictor

The reason is quite simple.
The different subtypes of cholesterol work together in an intricately balanced system.
There is a wide range of total cholesterol levels that are perfectly healthy, so long as the proportions of the subtypes are healthy ones.
By the same token, a given level of total cholesterol, even if it is perfectly normal, could be pathological when examined by subtype.
Strong evidence from recent decades suggests that the best known blood lipid measures for predicting future risk of CVD are HDL, triglycerides, and related ratios (see below).

Similarly, while LDL cholesterol is probably important, it appears that it does not have good predictive power when looking at its magnitude alone

One reason for this is that like total cholesterol, LDL is not uniform.
Just as we distinguish between HDL and LDL, the so-called “good” and “bad” cholesterol, LDL itself is now known to have two important subtypes with opposite risk
Having more large, light LDL particles (also called Pattern A), does not indicate high CVD risk, but having more small, dense particles (Pattern B) does
Therefore high LDL by itself is not necessarily indicative of CVD.

Low HDL cholesterol is strongly associated with CVD

Having high blood levels of HDL is now widely recognized as predicting lower levels of heart disease.
The proportion of total cholesterol that is HDL cholesterol is a particularly strong predictor.
In 2007, a meta-analysis was published in the Lancet that examined information from 61 prospective observational studies, consisting of almost 900,000 adults.
Information about HDL was available for about 150,000 of them, among whom there were 5000 vascular deaths. According to the authors, “the ratio of total to HDL cholesterol is a substantially more informative predictor of IHD mortality than are total cholesterol, HDL
cholesterol, or non-HDL cholesterol.”

This is consistent with many other studies, for example this very recent analysis from the COURAGE trial

High triglycerides are strongly associated with CVD

There has been drawn out controversy in the medical community as to the relationship of triglyceride levels to CVD.
There are two parts to the controversy: whether or not triglycerides are an independent predictor of CVD, and whether or not triglycerides play a causative role in CVD.

In both cases, however, it doesn’t matter in which way the controversy is resolved!
Whether or not triglycerides independently predict CVD
(and there is at least some evidence that they do),
and whether or not they cause CVD, there is no controversy about whether they predict CVD.
The association between triglyceride levels and CVD still holds and is strongly predictive
In fact it is so predictive that those who argue that triglyceride levels are not an independent risk factor, call it instead a “biomarker” for CVD
In other words, seeing high triglycerides is tantamount to seeing the progression of heart disease.

HDL-to-Triglycerides Ratio: compounding evidence

Triglycerides and HDL levels statistically interact.
That means it is a mistake to treat one as redundant with respect to the other.
If you do, you will miss the fact that the effect of one on your outcome of interest changes depending on the value of the other.
Despite the fact that most heart disease researchers who study risk factors have not used methods tuned to find interactions between triglycerides and HDL, many studies have at least measured both.
This has allowed others to do the appropriate analysis.
When triglycerides and HDL have been examined with respect to each other, that is, when the effect of triglycerides is measured under the condition of low HDL, or when the effect of HDL is measured under the condition of high triglycerides, this combination of factors turns out to be even more indicative of CVD

One of the most interesting aspects of this finding from our perspective, is that the ratio of triglyceride levels to HDL is considered to be a surrogate marker of insulin resistance
(See The Ketogenic Diet as a Treatment for Metabolic Syndrome.)
In other words, the best lipid predictors of CVD are also those that indicate insulin resistance.

Ketogenic Diets improve risk factors for CVD

There is now ample evidence that a low carbohydrate, ketogenic diet improves lipid profiles, particularly with respect to the risk factors outlined above: triglycerides, HDL, and their ratio

Although a ketogenic diet typically raises LDL levels, which has been traditionally seen as a risk factor, it has also been shown to improve LDL particle size.
In other words, although the absolute amount of LDL goes
up, it is the “good” LDL that goes up, whereas the “bad” LDL goes down
This is hardly surprising, since LDL particle size is also strongly predicted by triglycerides

Although there have not yet been intervention studies testing the effect of a ketogenic diet on the rate of actual CVD incidents (e.g. heart attacks), the evidence about lipid profiles is strong enough to make ketogenic diets more likely to reduce heart disease than any other known intervention.


  • Current medical practice uses blood lipid measurements to assess the risk of heart disease.
  • Despite the continuing tradition of measuring total cholesterol and LDL, we have known for decades that triglycerides, HDL, and the ratio of the two, are much better predictors of heart disease.
    LDL particle size is also considered strongly predictive.
  • A ketogenic diet has a very favourable impact on these risk factors, and thus should be considered the diet of choice for those at risk of CVD.

In their 2011 paper, “Low-carbohydrate diet review: shifting the paradigm”, Hite et al. display the following graph (VLCKD stands for Very Low Carbohydrate Ketogenic Diet, and LFD for Low Fat Diet)
36 based on data from 31:

It makes an excellent visualization of the factors at stake, and how powerful a ketogenic diet is.
It also shows quite clearly that not only is restricting carbohydrate more effective for this purpose than a low fat diet, but that a low fat diet is detrimental for some important risk factors — apolipoprotein ratios, LDL particle size, and HDL — but a low carb diet is not.
The ketogenic diet resulted in a significant improvement in every measure.


1 Evidence type: observational

World Health Organization Fact sheet N°317: Cardiovascular diseases (CVDs) September 2011

  • CVDs are the number one cause of death globally: more people die annually from CVDs than from any other cause.
  • An estimated 17.3 million people died from CVDs in 2008, representing 30% of all global deaths. Of these deaths, an estimated 7.3 million were due to coronary heart disease and 6.2 million were due to stroke.
  • Low- and middle-income countries are disproportionally affected: over 80% of CVD deaths take place in low- and middle-income countries and occur almost equally in men and women.
  • By 2030, almost 23.6 million people will die from CVDs, mainly from heart disease and stroke. These are projected to remain the single leading causes of death.

2 Evidence type: observational

Role of lipid and lipoprotein profiles in risk assessment and therapy.
Ballantyne CM, Hoogeveen RC.
Am Heart J. 2003 Aug;146(2):227-33.

Despite a strong and consistent association within populations, elevated TC [(total cholesterol)] alone is not a useful test to discriminate between individuals who will have CHD [(coronary heart disease)] events and those who will not.

3 Evidence type: observational

Relation of serum lipoprotein cholesterol levels to presence and severity of angiographic coronary artery disease.
Philip A. Romm, MD, Curtis E. Green, MD, Kathleen Reagan, MD, Charles E. Rackley, MD.
The American Journal of Cardiology Volume 67, Issue 6, 1 March 1991, Pages 479–483

Most CAD [(coronary artery disease)] occurs in persons who have only mild or moderate elevations in cholesterol levels. Total cholesterol level alone is a poor predictor of CAD, particularly in older patients in whom the major lipid risk factor is the HDL cholesterol level.

4 Evidence type: observational

Lipids, risk factors and ischaemic heart disease.
Atherosclerosis. 1996 Jul;124 Suppl:S1-9.
Castelli WP.

Those individuals who had TC [(total cholesterol)] levels of 150-300 mg/dl (3.9-7.8 mmol/1) fell into the overlapping area (Fig. 1), demonstrating that 90% of the TC levels measured were useless (by themselves) for predicting risk of CHD [(coronary heart disease)] in a general population. Indeed, twice as many individuals who had a lifetime TC level of less than 200 mg/dl (5.2 mmol/1) had CHD compared with those who had a TC level greater than 300 mg/dl (7.8 mmol/l) (Fig. 1).

5 Evidence type: observational

Range of Serum Cholesterol Values in the Population Developing Coronary Artery Disease.
William B. Kannel, MD, MPH.
The American Journal of Cardiology, Volume 76, Issue 9, Supplement 1, 28 September 1995, Pages 69C–77C

The ranges of serum cholesterol and LDL cholesterol levels varied widely both in the general population and in patients who had already manifested CAD (Figures 1 and 2). Because of the extensive overlap between levels, it was impossible to differentiate the patients with CAD from the control subjects.

6 Evidence type: observational

Lipoprotein cholesterol, apolipoprotein A-I and B and lipoprotein (a) abnormalities in men with premature coronary artery disease.
Jacques Genest Jr., MD,FACC, Judith R. McNamara, MT, Jose M. Ordovas, PhD, Jennifer L. Jenner, BSc, Steven R. Silberman, PhD, Keaven M. Anderson, PhD, Peter W.F. Wilson, MD, Deeb N. Salem, MD, FACC, Ernst J. Schaefer, MD.
Journal of the American College of Cardiology Volume 19, Issue 4, 15 March 1992, Pages 792–802.

Our data suggest that total and LDL cholesterol may not be the best discriminants for the presence of coronary artery disease despite the strong association between elevated cholesterol and the development of coronary artery disease in cross-sectional population studies and prospective epidemiologic studies.

7 Evidence type: observational

Apolipoprotein B and apolipoprotein A-I: risk indicators of coronary heart disease and targets for lipid-modifying therapy.
Walldius, G. and Jungner, I. (2004),
Journal of Internal Medicine, 255: 188–205. doi: 10.1046/j.1365-2796.2003.01276.x

(Emphasis ours.)

For over three decades it has been recognized that a high level of total blood cholesterol, particularly in the form of LDL cholesterol (LDL-C), is a major risk factor for developing coronary heart disease (CHD) [1–4]. However, as more recent research has expanded our understanding of lipoprotein function and metabolism, it has become apparent that LDL-C is not the only lipoprotein species involved in atherogenesis. A considerable proportion of patients with atherosclerotic disease have levels of LDL-C and total cholesterol (TC) within the recommended range [5, 6], and some patients who achieve significant LDL-C reduction with lipid-lowering therapy still develop CHD [7].

Other lipid parameters are also associated with elevated cardiovascular risk, and it has been suggested that LDL-C and TC may not be the best discriminants for the presence of coronary artery disease (CAD) [5].

8 Evidence type: observational

Plasma Lipoprotein Levels as Predictors of Cardiovascular Death in Women.
Katherine Miller Bass, MD, MHS; Craig J. Newschaffer, MS; Michael J. Klag, MD, MPH; Trudy L. Bush, PhD, MHS.
Arch Intern Med. 1993;153(19):2209-2216.

Using a sample of 1405 women aged 50 to 69 years from the Lipid Research Clinics’ Follow-up Study, age-adjusted CVD death rates and summary relative risk (RR) estimates by categories of lipid and lipoprotein levels were calculated. Multivariate analysis was performed to provide RR estimates adjusted for other CVD risk factors.

RESULTS: Average follow-up was 14 years. High-density lipoprotein and triglyceride levels were strong predictors of CVD death in age-adjusted and multivariate analyses. Low-density lipoprotein and total cholesterol levels were poorer predictors of CVD mortality. After adjustment for other CVD risk factors, HDL levels less than 1.30 mmol/L (50 mg/dL) were strongly associated with cardiovascular mortality (RR = 1.74; 95% confidence interval [CI], 1.10 to 2.75). Triglyceride levels were associated with increased CVD mortality at levels of 2.25 to 4.49 mmol/L (200 to 399 mg/dL) (RR = 1.65; 95% CI, 0.99 to 2.77) and 4.50 mmol/L (400 mg/dL) or greater (RR = 3.44; 95% CI, 1.65 to 7.20). At total cholesterol levels of 5.20 mmol/L (200 mg/dL) or greater and at all levels of LDL and triglycerides, women with HDL levels of less than 1.30 mmol/L (< 50 mg/dL) had CVD death rates that were higher than those of women with HDL levels of 1.30 mmol/L (50 mg/dL) or greater.

9 Evidence type: plausible mechanism and observational review

Particle size: the key to the atherogenic lipoprotein?
Rajman I, Maxwell S, Cramb R, Kendall M.
QJM. 1994 Dec;87(12):709-20.

Using different analytical methods, up to 12 low-density lipoprotein (LDL) subfractions can be separated. LDL particle size decreases with increasing density. Smaller, denser LDL particles seem more atherogenic than the larger, lighter particles, based on the experimental findings that smaller LDL particles are more susceptible for oxidation in vitro, have lower binding affinity for the LDL receptors and lower catabolic rate, have a higher concentration of polyunsaturated fatty acids, and potentially interact more easily with proteoglycans of the arterial wall. Clinical studies have shown that a smaller LDL subfraction profile is associated with an increased risk of heart disease, even when total cholesterol level is only slightly raised. There is a strong inverse association between LDL particle size and triglyceride concentrations. Although LDL particle size is genetically determined, its phenotypic expression may also be affected by environmental factors such as drugs, diet, obesity, exercise or disease. Factors that shift the LDL subfractions profile towards larger particles may reduce the risk of heart disease.

10 Evidence type: nested case-control study

Association of Small Low-Density Lipoprotein Particles With the Incidence of Coronary Artery Disease in Men and Women.
Christopher D. Gardner, PhD; Stephen P. Fortmann, MD; Ronald M. Krauss, MD
JAMA. 1996;276(11):875-881. doi:10.1001/jama.1996.03540110029028.

Incident CAD cases were identified through FCP surveillance between 1979 and 1992. Controls were matched by sex, 5-year age groups, survey time point, ethnicity, and FCP treatment condition. The sample included 124 matched pairs: 90 pairs of men and 34 pairs of women.

LDL size was smaller among CAD cases than controls (mean ±SD) (26.17±1.00nm vs 26.68±0.90nm;P

The Ketogenic Diet as a Treatment for Metabolic Syndrome

sup {font-size: .6em;}

The Ketogenic Diet for Metabolic Syndrome

Metabolic Syndrome (MetS) can be viewed as a set of symptoms of insulin resistance.
Taken together, those symptoms signify a threat of heart disease, diabetes, cancer, and other diseases that appear to be different manifestations of a common cause.
That common cause is likely to be insulin resistance.

This hypothesis is supported by evidence that ketogenic diets not only normalize insulin sensitivity and the symptoms of MetS, but they treat (or have promise in treating) many MetS-associated diseases.

In light of this, it seems plausible that adopting a ketogenic diet will significantly improve your chances of avoiding these diseases in the first place.

* * *

In brief

  • Metabolic Syndrome is a cluster of symptoms, not a disease.
    Those symptoms are useful to class together, because their association with a variety of different diseases strongly suggests a common cause.
    In other words, it has provided us with a compelling hypothesis.
  • If there were a common cause, then a therapy that treats that cause should help them all.
    Moreover, it should reduce the symptoms of Metabolic Syndrome itself.
    Further, treatments that work for one but not the others should be considered inferior, “band-aid” treatments.
  • A ketogenic diet improves Metabolic Syndrome.
    Also, for every disease associated with MetS that we have investigated, a keto diet has either been shown to help, has shown preliminary evidence in its favour, or has not been sufficiently tested to rule out.
  • This supports the hypothesis that those diseases have a common cause, and that a ketogenic diet addresses it.

* * *

What is Metabolic Syndrome?

Metabolic Syndrome is a cluster of symptoms that commonly occur together and indicate increased risk of cardiovascular disease (CVD), type 2 diabetes (T2D), cancer, and other diseases.
Clinically, to be diagnosed with MetS, you have to score above (or in the case of HDL, below) a healthy threshold in at least 3 of the following 5 measurements: waist size, fasting blood glucose, blood pressure, triglycerides, and HDL.
All of these are associated with insulin resistance, although some are more predictive than others
, and so metabolic syndrome might be more accurately described as insulin resistance syndrome (and it sometimes is)

Just as with any such measure, it can be misleading to draw a threshold at such a particular point.
The cost of ignoring warning signs because they fall below a threshold may be worse than the benefit of giving a special diagnosis to those who have multiple symptoms, each of which could be recognized as warranting treatment on its own

Nonetheless, it is useful to have a name for a set of associations for two reasons.

  1. It allows us to recognize the commonalities in symptoms of a variety of disease states which is suggestive of common mechanisms.
  2. It promotes the insight that any treatment that is purported to improve risk of CVD or T2D ought to have a beneficial impact on all of the associated symptoms.
    If it doesn’t, there is a risk that it is a band-aid solution that temporarily hides the problem rather than fixing it.

Because these symptoms so often occur together, and because they are all risk factors for a group of diseases which in turn are risk factors for each other, it is the contention of many scientists that they have a common cause.
Some argue that this common cause is obesity itself.
A separate cause is postulated for obesity, which then is supposed to cause the other risk factors.
However, other researchers, ourselves among them, believe that obesity and the other symptoms have a common cause related to insulin signalling.
For this reason, we have grouped together several diseases which appear to have insulin signalling at their root, and which have elevated risk in the presence of Metabolic Syndrome symptoms.
These diseases include (but are not limited to) cardiovascular disease
type 2 diabetes
polycystic ovarian syndrome
Alzheimer’s disease
10, and cancer.

In other words, we believe that Metabolic Syndrome is not itself a disease, but is a class of warning signs associated with the progression of several other diseases. If this is true, then when you treat the underlying cause of these symptoms, they will all normalize together, and the risk of all associated diseases will simultaneously be reduced.

* * *

Ketogenic diets treat insulin resistance and therefore are expected to treat all diseases that have Metabolic Syndrome as a symptom.

The following is just a sample of evidence showing that not only does a keto diet address the symptoms of MetS itself, but also those conditions associated with it.
This is not meant to be comprehensive — there are many more supporting experiments in each category!

  • Carbohydrate restriction has a more favorable impact on the metabolic syndrome than a low fat diet12.
  • A ketogenic diet favorably affects serum biomarkers for cardiovascular disease in normal-weight men
  • In addition to decreasing body weight and improving glycemia, a ketogenic diet can be effective in decreasing antidiabetic medication dosage
    14 .
  • In a pilot study, a ketogenic diet led to significant improvement in weight, percent free testosterone, LH/FSH ratio, and fasting insulin in women with obesity and PCOS over a 24 week period
  • An oral ketogenic compound, AC-1202, was tested in subjects with probable Alzheimer’s disease, and resulted in a significant improvement to cognitive scores
  • It seems a reasonable possibility that a very-low-carbohydrate diet could help to reduce the progression of some types of cancer, although at present the evidence is preliminary

* * *


  • The ketogenic diet is a powerful therapy that exerts its healing effect in a wide variety of conditions that may seem superficially unrelated.
  • These conditions are linked by their connection to insulin resistance, and therefore their association with MetS.
  • This supports not only the hypothesis that a keto diet treats MetS, but also that insulin resistance is the underlying cause of many devastating diseases, and that the way a keto diet is treating those is
    by intercepting and correcting the underlying cause.

* * *


1 Evidence type: observational analysis
Evidence type:

Insulin resistance in aging is related to abdominal obesity.
Kohrt WM, Kirwan JP, Staten MA, Bourey RE, King DS, Holloszy JO.
Diabetes. 1993 Feb;42(2):273-81.

(emphasis ours)


Studies have shown that insulin resistance increases with age, independent of changes in total adiposity. However, there is growing evidence that the development of insulin resistance may be more closely related to abdominal adiposity. To evaluate the independent effects of aging and regional and total adiposity on insulin resistance, we performed hyperinsulinemic euglycemic clamps on 17 young (21-33 yr) and 67 older (60-72 yr) men and women. We assessed FFM and total and regional adiposity by hydrodensitometry and anthropometry. Insulin-stimulated GDRs at a plasma insulin concentration of approximately 450 pM averaged 45.6 +/- 3.3 FFM-1 x min-1 (mean +/- SE) in the young subjects, 45.6 +/- 10.0 FFM-1 x min-1 in 24 older subjects who were insulin sensitive, and 23.9 +/- 11.7 FFM-1 x min-1 in 43 older subjects who were insulin resistant. Few significant differences were apparent in skin-fold and circumference measurements between young and insulin-sensitive older subjects, but measurements at most central body sites were significantly larger in the insulin-resistant older subjects. Waist girth accounted for > 40% of the variance in insulin action, whereas age explained only 10-20% of the total variance and < 2% of the variance when the effects of waist circumference were statistically controlled. These results suggest that insulin resistance is more closely associated with abdominal adiposity than with age."]

2 Evidence type: retrospective observation

Use of waist circumference to predict insulin resistance: retrospective study.
Wahrenberg H, Hertel K, Leijonhufvud BM, Persson LG, Toft E, Arner P.
BMJ. 2005 Jun 11;330(7504):1363-4. Epub 2005 Apr 15.

In the multiple regression model, waist circumference was the strongest regressor of the five significant covariates (standardised partial regression coefficients: waist circumference β1 = 0.37; log-plasma triglycerides β2 = 0.23; systolic blood pressure β3 = 0.10, high density lipoprotein cholesterol β4 = -0.09; and body mass index β5 = 0.15 (P < 0.001)).

3 Evidence type: observational analysis

Biomarkers in Fasting Serum to Estimate Glucose Tolerance, Insulin Sensitivity, and Insulin Secretion
Allison B. Goldfine, Robert W. Gerwien, Janice A. Kolberg, Sheila O’Shea, Sarah Hamren, Glenn P. Hein, Xiaomei M. Xu, and Mary Elizabeth Patti
Clinical Chemistry 57:2 326–337 (2011)

A subset of 5 markers was associated with insulin sensitivity (assessed using the dynamic CISI measure): fasting glucose, insulin, Fas ligand, complement C3, and PAI-1. As shown in Fig. 3C, 91% of variance between predicted and observed CISI values was accounted for by these 5 markers alone (P 0.0001). In addition, a bootstrap R 2 value of 0.90 (IQR 0.83–0.94) indicates that the model could be expected to perform well on an independent data set. By comparison, HOMA-IR, a widely accepted estimate of insulin resistance based on fasting glucose and insulin, explained 88% of the variance of the dynamic measure of insulin sensitivity.

4 Evidence type: observation

A.D.A.M. Medical Encyclopedia.

Metabolic syndrome; Insulin resistance syndrome; Syndrome X

5 Evidence type: observation

Diabetes Health Center Insulin Resistance and Diabetes

If you have pre-diabetes or diabetes, chances are that you’ve heard of the medical term insulin resistance syndrome or metabolic syndrome. Insulin resistance or metabolic syndrome describes a combination of health problems that have a common link — an increased risk of diabetes and early heart disease.

6 Evidence type: observation

The metabolic syndrome: is this diagnosis necessary?
Gerald M Reaven.
Am J Clin Nutr June 2006 vol. 83 no. 6 1237-1247

The goal of diagnosing the metabolic syndrome is to identify persons at increased risk of CVD. Because each component that makes up the versions of the metabolic syndrome increases CVD risk (34, 36, 37, 62, 68, 69), it seems prudent to treat any of these abnormalities that are present. Furthermore, it would not be too surprising that the more abnormalities present in any given person, the greater would be his or her risk of CVD. The question can be raised, however, as to whether identifying a person as having metabolic syndrome necessarily indicates that he or she is at greater risk of CVD than is a person who may not qualify for that designation. This did not seem to be the case when the ATP III criteria were applied to the Framingham Study database (117); a recent report pointed out that persons meeting any 2 criteria were at no less risk than were those meeting 3 criteria. Indeed, it would be possible to describe a number of prototypic clinical situations in which a person with 1 or 2 abnormalities would be at greater risk of CVD than would a patient who met the metabolic syndrome diagnostic criteria.

7 Evidence type: retrospective observation

The Metabolic Syndrome and Total and Cardiovascular Disease Mortality in Middle-aged Men.
Hanna-Maaria Lakka, MD, PhD; David E. Laaksonen, MD, MPH; Timo A. Lakka, MD, PhD; Leo K. Niskanen, MD, PhD; Esko Kumpusalo, MD, PhD; Jaakko Tuomilehto, MD, PhD; Jukka T. Salonen, MD, PhD
JAMA. 2002;288(21):2709-2716. doi:10.1001/jama.288.21.2709.

The metabolic syndrome, a concurrence of disturbed glucose and insulin metabolism, overweight and abdominal fat distribution, mild dyslipidemia, and hypertension, is associated with subsequent development of type 2 diabetes mellitus and cardiovascular disease (CVD).

The prevalence of the metabolic syndrome ranged from 8.8% to 14.3%, depending on the definition. There were 109 deaths during the approximately 11.4-year follow-up, of which 46 and 27 were due to CVD and CHD, respectively. Men with the metabolic syndrome as defined by the NCEP were 2.9 (95% confidence interval [CI], 1.2-7.2) to 4.2 (95% CI, 1.6-10.8) times more likely and, as defined by the WHO, 2.9 (95% CI, 1.2-6.8) to 3.3 (95% CI, 1.4-7.7) times more likely to die of CHD after adjustment for conventional cardiovascular risk factors. The metabolic syndrome as defined by the WHO was associated with 2.6 (95% CI, 1.4-5.1) to 3.0 (95% CI, 1.5-5.7) times higher CVD mortality and 1.9 (95% CI, 1.2-3.0) to 2.1 (95% CI, 1.3-3.3) times higher all-cause mortality. The NCEP definition less consistently predicted CVD and all-cause mortality. Factor analysis using 13 variables associated with metabolic or cardiovascular risk yielded a metabolic syndrome factor that explained 18% of total variance. Men with loadings on the metabolic factor in the highest quarter were 3.6 (95% CI, 1.7-7.9), 3.2 (95% CI, 1.7-5.8), and 2.3 (95% CI, 1.5-3.4) times more likely to die of CHD, CVD, and any cause, respectively.

Cardiovascular disease and all-cause mortality are increased in men with the metabolic syndrome, even in the absence of baseline CVD and diabetes.

8 Evidence type: retrospective observation

Risks for All-Cause Mortality, Cardiovascular Disease, and Diabetes Associated With the Metabolic Syndrome: A summary of the evidence.
Earl S. Ford, MD, MPH
Diabetes Care July 2005 vol. 28 no. 7 1769-1778

For studies that used the exact NCEP definition of the metabolic syndrome, random-effects estimates of combined relative risk were 1.27 (95% CI 0.90–1.78) for all-cause mortality, 1.65 (1.38–1.99) for cardiovascular disease, and 2.99 (1.96–4.57) for diabetes. For studies that used the most exact WHO definition of the metabolic syndrome, the fixed-effects estimates of relative risk were 1.37 (1.09–1.74) for all-cause mortality and 1.93 (1.39–2.67) for cardiovascular disease; the fixed-effects estimate was 2.60 (1.55–4.38) for coronary heart disease.

CONCLUSIONS—These estimates suggest that the population-attributable fraction for the metabolic syndrome, as it is currently conceived, is ∼6–7% for all-cause mortality, 12–17% for cardiovascular disease, and 30–52% for diabetes.

9 Evidence type: retrospective observation

Prevalence and Characteristics of the Metabolic Syndrome in Women with Polycystic Ovary Syndrome.
Teimuraz Apridonidze, Paulina A. Essah, Maria J. Iuorno and John E. Nestler.
The Journal of Clinical Endocrinology & Metabolism April 1, 2005 vol. 90 no. 4 1929-1935

The polycystic ovary syndrome (PCOS) is characterized by insulin resistance with compensatory hyperinsulinemia. Insulin resistance also plays a role in the metabolic syndrome (MBS). We hypothesized that the MBS is prevalent in PCOS and that women with both conditions would present with more hyperandrogenism and menstrual cycle irregularity than women with PCOS only.

We conducted a retrospective chart review of all women with PCOS seen over a 3-yr period at an endocrinology clinic. Of the 161 PCOS cases reviewed, 106 met the inclusion criteria. The women were divided into two groups: 1) women with PCOS and the MBS (n = 46); and 2) women with PCOS lacking the MBS (n = 60).

Prevalence of the MBS was 43%, nearly 2-fold higher than that reported for age-matched women in the general population. Women with PCOS had persistently higher prevalence rates of the MBS than women in the general population, regardless of matched age and body mass index ranges.

10 Evidence type: retrospective observation

Association of metabolic syndrome with Alzheimer disease: A population-based study.
M. Vanhanen, PhD, K. Koivisto, MD, PhD, L. Moilanen, MD, PhD, E. L. Helkala, PhD, T. Hänninen, PhD, H. Soininen, MD, PhD, K. Kervinen, MD, PhD, Y. A. Kesäniemi, MD, PhD, M. Laakso, MD, PhD and J. Kuusisto, MD, PhD
Neurology September 12, 2006 vol. 67 no. 5 843-847

Of the study subjects, 418 (43.6%) had MetS. Probable or possible AD was diagnosed in 45 subjects (4.7%). AD was more frequently detected in subjects with MetS than in subjects without MetS (7.2 vs 2.8%; p < 0.001). The prevalence of AD was higher in women with MetS vs women without the syndrome (8.3 vs 1.9%; p < 0.001), but in men with MetS, the prevalence of AD was not increased (3.8 vs 3.9%; p = 0.994). In univariate logistic regression analysis, MetS was significantly associated with AD (odds ratio [OR] 2.71; 95% CI 1.44 to 5.10). In multivariate logistic regression analysis including also apolipoprotein E4 phenotype, education, age, and total cholesterol, MetS was significantly associated with AD (OR 2.46; 95% CI 1.27 to 4.78). If only nondiabetic subjects were included in the multivariate analysis, MetS was still significantly associated with AD (OR 3.26; 95% CI 1.45 to 7.27).

11 Evidence type: review and meta-analysis

Metabolic syndrome and risk of cancer: a systematic review and meta-analysis.
Esposito K, Chiodini P, Colao A, Lenzi A, Giugliano D.
Diabetes Care. 2012 Nov;35(11):2402-11. doi: 10.2337/dc12-0336.

RESULTS: We analyzed 116 datasets from 43 articles, including 38,940 cases of cancer. In cohort studies in men, the presence of metabolic syndrome was associated with liver (relative risk 1.43, P < 0.0001), colorectal (1.25, P < 0.001), and bladder cancer (1.10, P = 0.013). In cohort studies in women, the presence of metabolic syndrome was associated with endometrial (1.61, P = 0.001), pancreatic (1.58, P < 0.0001), breast postmenopausal (1.56, P = 0.017), rectal (1.52, P = 0.005), and colorectal (1.34, P = 0.006) cancers. Associations with metabolic syndrome were stronger in women than in men for pancreatic (P = 0.01) and rectal (P = 0.01) cancers. Associations were different between ethnic groups: we recorded stronger associations in Asia populations for liver cancer (P = 0.002), in European populations for colorectal cancer in women (P = 0.004), and in U.S. populations (whites) for prostate cancer (P = 0.001). CONCLUSIONS: Metabolic syndrome is associated with increased risk of common cancers; for some cancers, the risk differs betweens sexes, populations, and definitions of metabolic syndrome.

12 Evidence type: controlled experiment

Carbohydrate restriction has a more favorable impact on the metabolic syndrome than a low fat diet.
Volek JS, Phinney SD, Forsythe CE, Quann EE, Wood RJ, Puglisi MJ, Kraemer WJ, Bibus DM, Fernandez ML, Feinman RD.
Lipids. 2009 Apr;44(4):297-309. doi: 10.1007/s11745-008-3274-2. Epub 2008 Dec 12.


We recently proposed that the biological markers improved by carbohydrate restriction were precisely those that define the metabolic syndrome (MetS), and that the common thread was regulation of insulin as a control element. We specifically tested the idea with a 12-week study comparing two hypocaloric diets (approximately 1,500 kcal): a carbohydrate-restricted diet (CRD) (%carbohydrate:fat:protein = 12:59:28) and a low-fat diet (LFD) (56:24:20) in 40 subjects with atherogenic dyslipidemia. Both interventions led to improvements in several metabolic markers, but subjects following the CRD had consistently reduced glucose (-12%) and insulin (-50%) concentrations, insulin sensitivity (-55%), weight loss (-10%), decreased adiposity (-14%), and more favorable triacylglycerol (TAG) (-51%), HDL-C (13%) and total cholesterol/HDL-C ratio (-14%) responses. In addition to these markers for MetS, the CRD subjects showed more favorable responses to alternative indicators of cardiovascular risk: postprandial lipemia (-47%), the Apo B/Apo A-1 ratio (-16%), and LDL particle distribution. Despite a threefold higher intake of dietary saturated fat during the CRD, saturated fatty acids in TAG and cholesteryl ester were significantly decreased, as was palmitoleic acid (16:1n-7), an endogenous marker of lipogenesis, compared to subjects consuming the LFD. Serum retinol binding protein 4 has been linked to insulin-resistant states, and only the CRD decreased this marker (-20%). The findings provide support for unifying the disparate markers of MetS and for the proposed intimate connection with dietary carbohydrate. The results support the use of dietary carbohydrate restriction as an effective approach to improve features of MetS and cardiovascular risk.

13 Evidence type: non-randomized experiment

A Ketogenic Diet Favorably Affects Serum Biomarkers for Cardiovascular Disease in Normal-Weight Men.
Matthew J. Sharman, William J. Kraemer, Dawn M. Love, Neva G. Avery, Ana L. Gómez, Timothy P. Scheett, and Jeff S. Volek.
J. Nutr. July 1, 2002 vol. 132 no. 7 1879-1885

The primary objective of this study was to examine how healthy normolipidemic, normal-weight men respond to a ketogenic diet in terms of fasting and postprandial CVD biomarkers. Ketogenic diets have been criticized on the grounds they jeopardize health (8); however, very few studies have directly evaluated the effects of a ketogenic diet on fasting and postprandial risk factors for CVD. Subjects consumed a diet that consisted of 8% carbohydrate (27 kg/m2 and a clinical diagnosis of PCOS were recruited from the community. They were instructed to limit their carbohydrate intake to 20 grams or less per day for 24 weeks. Participants returned every two weeks to an outpatient research clinic for measurements and reinforcement of dietary instruction. In the 5 women who completed the study, there were significant reductions from baseline to 24 weeks in body weight (-12%), percent free testosterone (-22%), LH/FSH ratio (-36%), and fasting insulin (-54%). There were non-significant decreases in insulin, glucose, testosterone, HgbA1c, triglyceride, and perceived body hair. Two women became pregnant despite previous infertility problems.

16 Evidence type: randomized, double-blind, placebo-controlled, multicenter trial

Study of the ketogenic agent AC-1202 in mild to moderate Alzheimer’s disease: a randomized, double-blind, placebo-controlled, multicenter trial.
Samuel T Henderson, Janet L Vogel, Linda J Barr, Fiona Garvin, Julie J Jones and Lauren C Costantini.
Nutrition & Metabolism 2009, 6:31

AC-1202 significantly elevated a serum ketone body (β-hydroxybutyrate) 2 hours after administration when compared to Placebo. In each of the population groups, a significant difference was found between AC-1202 and Placebo in mean change from Baseline in ADAS-Cog score on Day 45: 1.9 point difference, p = 0.0235 in ITT; 2.53 point difference, p = 0.0324 in per protocol; 2.6 point difference, p = 0.0215 in dosage compliant. Among participants who did not carry the APOE4 allele (E4(-)), a significant difference was found between AC-1202 and Placebo in mean change from Baseline in ADAS-Cog score on Day 45 and Day 90. In the ITT population, E4(-) participants (N = 55) administered AC-1202 had a significant 4.77 point difference in mean change from Baseline in ADAS-Cog scores at Day 45 (p = 0.0005) and a 3.36 point difference at Day 90 (p = 0.0148) compared to Placebo. In the per protocol population, E4(-) participants receiving AC-1202 (N = 37) differed from placebo by 5.73 points at Day 45 (p = 0.0027) and by 4.39 points at Day 90 (p = 0.0143). In the dosage compliant population, E4(-) participants receiving AC-1202 differed from placebo by 6.26 points at Day 45 (p = 0.0011, N = 38) and 5.33 points at Day 90 (p = 0.0063, N = 35). Furthermore, a significant pharmacologic response was observed between serum β-hydroxybutyrate levels and change in ADAS-Cog scores in E4(-) subjects at Day 90 (p = 0.008).

17 Evidence type: review of experiments and case-studies

Beyond weight loss: a review of the therapeutic uses of very-low-carbohydrate (ketogenic) diets.
A Paoli, A Rubini, J S Volek and K A Grimaldi.
European Journal of Clinical Nutrition (2013) 67, 789–796; doi:10.1038/ejcn.2013.116; published online 26 June 2013

[I]t seems a reasonable possibility that a very-low-carbohydrate diet could help to reduce the progression of some types of cancer, although at present the evidence is preliminary. In the 1980s, seminal animal studies by Tisdale and colleagues demonstrated that a ketogenic diet was capable to reduce tumour size in mice, whereas more recent research has provided evidence that ketogenic diets may reduce tumour progression in humans, at least as far as gastric and brain cancers are concerned. Although no randomized controlled trials with VLCKD have yet been conducted on patients and the bulk of evidence in relation to the influence of VLCKD on patient survival is still anecdotal, a very recent paper by Fine et al. suggests that the insulin inhibition caused by a ketogenic diet could be a feasible adjunctive treatment for patients with cancer. In summary, perhaps through glucose ‘starvation’ of tumour cells and by reducing the effect of direct insulin-related actions on cell growth, ketogenic diets show promise as an aid in at least some kind of cancer therapy and is deserving of further and deeper investigation—certainly the evidence justifies setting up clinical trials.

The medical-grade diet

sup {font-size: .6em;}

This post is based on a talk we gave at BSidesLV on August 1st, 2013. You can also watch the video (20 minutes long).

In the face of a severe medical condition, typical dietary therapies have little or no power.
The ketogenic diet, however, has proven potency.
Here’s how it’s different.

In brief

  • Ketogenic diets are uniquely powerful among dietary therapies.
    They have been shown to have profound medical effects on serious conditions, especially in the brain.
  • These effects may be beneficial even if you aren’t currently sick.
  • This power is accessible to practically anyone.

Diets are weak medicine

If someone tells you they know of a diet that treats a serious medical condition, you ought to be skeptical.

We don’t mean a diet that restricts something you are allergic to or intolerant of — yes, we can prevent retardation in phenylketonurics by avoiding phenylalanine, and if you have gluten intolerance, avoiding gluten is a no-brainer.

However, if someone tells you that you can follow their diet and stop taking medication for a formidable disease like bipolar disorder, or that it would cure a progressive terminal disease like Alzheimer’s or cancer, you would probably think that person was a quack.
That’s because dietary therapies for major diseases don’t typically withstand scientific scrutiny.
When tested in clinical trials, they show no significant results, or weak results at best.

Diets might help you feel better or might slightly improve your chances, but they are unlikely to cause the sort of profound changes in your body that can completely reverse the course of a major disease.

The ketogenic diet is an exception.

Ketogenic diets are strong medicine

The ketogenic diet is currently used by doctors to treat a major illness, not as an “alternative therapy” — but as a standard, proven medical practice.
It is now recognized by neurologists that ketogenic diets are at least as effective as the best anti-epileptic drugs.
Around 15% of epilepsy patients who are put on a keto diet by their neurologist become completely seizure free. About a third have a 90% reduction in seizures, and a third have better than a 50% reduction in seizures&nbsp1.
That’s state-of-the-art treatment for that disease.

A keto diet induces a distinct, favourable metabolic state

As we described in our article on keto-adaptation, a ketogenic diet shifts your metabolism from relying mostly on glucose for fuel, to relying mostly on fuels derived from fat (including ketone bodies — a fuel used by all human bodies, but used in greater quantities when on a keto diet).
This change has extensive effects at both the cellular level, and the whole-body level.

It causes profound effects in the brain.
Although the mechanisms aren’t fully understood, it is well-established that ketogenic diets are neuroprotective against a variety of insults.
That is, they protect against different kinds of brain damage.
For example, in animal experiments, those animals on a ketogenic diet sustain significantly less brain damage after artificially induced stroke&nbsp2 or trauma&nbsp3.
It even protects against damage from nerve gas&nbsp4.

Ketogenic diets also improve conditions associated with heart disease and diabetes, such as high triglycerides and low HDL (“good cholesterol”)&nbsp5, insulin resistance&nbsp6, and obesity.&nbsp7

Moreover, there is preliminary evidence that ketogenic diets are effective against other serious conditions, including bipolar disorder&nbsp8, Alzheimer’s disease&nbsp9, and cancer&nbsp10.

Finally, on a keto diet your energy supply is managed more directly by your body.
So it makes sense that energy, focus, and mood would become more stable.
This has been reported anecdotally and we’ve previously given a possible explanation for it.

Try this at home!

Ketogenic diets are safe.
Like starting an exercise program, starting a keto diet doesn’t require a prescription or the supervision of a doctor, except in special cases&nbspfootnote †.
It’s not difficult to do (though there are some common pitfalls).
The potential benefits are high and the evidence for those benefits is strong.
As we argue in How to Judge a Health Practice, this is a rare and valuable combination of qualities.


  • Ketogenic diets are recognized as being at least as powerful as drugs in at least one major clinical condition, and there is reason to believe they are useful against others.
  • The effects may be beneficial even if you currently consider yourself to be healthy.
    Perhaps you could be even more healthy!
  • It’s easy and practical to try a ketogenic diet yourself.

footnote †
Like starting an exercise program, starting a keto diet is safe for healthy people. However, if you have a dangerous or unstable medical condition—for example you are in danger of having a heart attack or you have bipolar disorder—you should of course consult your doctor before doing something like starting an exercise program or a keto diet.
Also, of course, you should always consult a doctor before discontinuing a medication or changing your dosage. This is particularly relevant to keto diet, because for some conditions the diet could cause you to need less of your medication, such as for blood pressure, diabetes, epilepsy, or bipolar. Therefore, if you go on a keto diet and keep taking the same dosage of your medication, then you may end up taking too much medication for your needs, which itself could be dangerous. On the other hand, reducing or discontinuing your medication could be dangerous. Therefore, if you are currently taking a medication, you should consult your doctor before starting a keto diet.
This is not because keto diet is unsafe! At least, no more so than exercising. It’s just that if you have a dangerous or unstable medical condition or you are taking prescription medications, then you have to be extra careful about making changes.


1 Evidence type: review of clinical reports

Neal EG, Cross JH.
Efficacy of dietary treatments for epilepsy.
J Hum Nutr Diet. 2010 Apr;23(2):113-9. doi: 10.1111/j.1365-277X.2010.01043.x.

There have been two systematic reviews on the efficacy of the KD. The first included 11 studies published since 1970 (Lefevre & Aronson, 2000). Using a combined analysis of outcome data, the authors reported 15.8% of children to be seizure free and 55.8% to have a greater than 50% reduction in seizures, and it was concluded that there was sufficient evidence to determine that the diet is efficacious in children with intractable epilepsy. A more recent review of 14 studies arrived at the same conclusion (Keene, 2006), reporting an average of 15.6% of a total collective population of 972 patients to be seizure free after 6 months, with 33% having a greater than 50% seizure reduction. A statistical meta-analysis of 19 studies and a total of 1084 patients (Henderson et al., 2006) found the diet reduced seizures by >90% in one-third of the patients, regardless of age or seizure type; the pooled odds ratio of treatment success among patients staying on the diet relative to those discontinuing was 2.25.

2 Evidence type: review of controlled animal experiments

Carl E. Stafstrom and Jong M. Rho.
The Ketogenic Diet as a Treatment Paradigm for Diverse Neurological Disorders
Front Pharmacol. 2012; 3: 59.
Published online 2012 April 9. Prepublished online 2012 January 25. doi: 10.3389/fphar.2012.00059

To date, no clinical trials of the KD have been performed in patients with stroke, but several animal studies of hypoxia-ischemia support the potential beneficial effect of the diet. Most of these models entail pre-treatment with the KD (or with BHB), resulting in decreased structural and functional damage from the stroke.

3 Evidence type: review of controlled animal experiments

Carl E. Stafstrom and Jong M. Rho.
The Ketogenic Diet as a Treatment Paradigm for Diverse Neurological Disorders
Front Pharmacol. 2012; 3: 59.
Published online 2012 April 9. Prepublished online 2012 January 25. doi: 10.3389/fphar.2012.00059

Several recent animal studies support this idea [that dietary therapy might ameliorate brain injury and possibly, long-term consequences such as epilepsy], and investigators have principally focused on ketone bodies (Prins, 2008a). Using a controlled cortical impact (CCI) injury model, Prins et al. (2005) showed that pre-treatment with a KD significantly reduced cortical contusion volume in an age-related manner that correlated with maturation-dependent differences in cerebral metabolism and ketone utilization. Later, they showed that cognitive and motor functioning was also improved with KD treatment (Appelberg et al., 2009). Further, using a weight drop model, Hu et al. (2009) showed that the KD pre-treatment reduced Bcl-2 (also known as Bax) mRNA and protein levels 72 h after trauma, indicating that apoptotic neurodegeneration could be prevented with this diet. Consistent with these observations, it was found that fasting – which shares the key feature of ketosis with the KD – led to significant tissue sparing in brain following CCI injury, and that again ketosis (with improved mitochondrial functioning) rather than the relative hypoglycemia seen with fasting was the important determinant of neuroprotection (Davis et al., 2008).

4 Evidence type: controlled animal experiments

Jeffrey L. Langston, Todd M. Myers
Diet composition modifies the toxicity of repeated soman exposure in rats.
Neurotoxicology. 2011 Jun;32(3):342-9. doi: 10.1016/j.neuro.2011.03.001. Epub 2011 Mar 17.

Differences in toxicity as a function of diet composition became apparent during the first week. Specifically, rats fed the glucose-enriched diet showed pronounced intoxication during Week 1, resulting in imperfect survival, weight loss, and deteriorated avoidance performance relative to all other groups. All rats fed the glucose-enriched diet died by the end of exposure Week 2. In contrast, only 10% of animals fed the standard diet died by the end of Week 2. Also in Week 2, weight loss and disrupted avoidance performance were apparent for all groups except for those fed the ketogenic diet. This differential effect of diet composition became even more striking in Week 3 when survival in the standard and choline diet groups approximated 50%, whereas survival equaled 90% in the ketogenic diet group.

5 Evidence type: controlled clinical trial and review of past trials

Westman EC, Yancy WS Jr, Olsen MK, Dudley T, Guyton JR.
Effect of a low-carbohydrate, ketogenic diet program compared to a low-fat diet on fasting lipoprotein subclasses.
Int J Cardiol. 2006 Jun 16;110(2):212-6. Epub 2005 Nov 16.

Recent research implicates dietary carbohydrates, especially refined carbohydrates, as a risk factor for cardiovascular disease [1]. In recent studies, a low-carbohydrate, ketogenic diet (LCKD) led to weight loss and improvements in high-density lipoprotein cholesterol (HDL-C) and serum triglyceride over a 6- to 12-month period [2], [3], [4] and [5]. Because obesity, low HDL-C, and elevated triglyceride are recognized as cardiovascular risk factors, and can be made worse by a low-fat/high-carbohydrate diet [6], [7] and [8], an LCKD might be a candidate treatment for these conditions.

Triglyceride is increasingly thought to be important in the pathogenesis of atherosclerosis, and treatments that lower triglyceride and raise HDL-cholesterol have been shown to reduce major coronary events [38] and [39]. High triglyceride levels promote the formation of small LDL by a process of cholesterol ester/triglyceride exchange and subsequent lipase action on triglyceride-enriched LDL [40]. Small HDL appear to be formed by a similar process and are then cleared from the circulation more rapidly than large HDL [41]. Based on the findings of this study, the reduction of dietary carbohydrate should be evaluated as a treatment for hypertriglyceridemia, low HDL-C, and ultimately atherosclerosis.

6 Evidence type: small clinical trial

Boden G, Sargrad K, Homko C, Mozzoli M, Stein TP.
Effect of a low-carbohydrate diet on appetite, blood glucose levels, and insulin resistance in obese patients with type 2 diabetes.
Ann Intern Med. 2005 Mar 15;142(6):403-11.

(Emphasis ours)

On the low-carbohydrate diet, mean energy intake decreased from 3111 kcal/d to 2164 kcal/d. The mean energy deficit of 1027 kcal/d (median, 737 kcal/d) completely accounted for the weight loss of 1.65 kg in 14 days (median, 1.34 kg in 14 days). Mean 24-hour plasma profiles of glucose levels normalized, mean hemoglobin A1c decreased from 7.3% to 6.8%, and insulin sensitivity improved by approximately 75%. Mean plasma triglyceride and cholesterol levels decreased (change, -35% and -10%, respectively).

7 Evidence type: meta-study of controlled trials

Hession M, Rolland C, Kulkarni U, Wise A, Broom J.
Systematic review of randomized controlled trials of low-carbohydrate vs. low-fat/low-calorie diets in the management of obesity and its comorbidities.
Obes Rev. 2009 Jan;10(1):36-50. doi: 10.1111/j.1467-789X.2008.00518.x. Epub 2008 Aug 11.

(Emphasis ours)
Note that these studies were on low carb diets that were not even necessarily so low as to be ketogenic.


There are few studies comparing the effects of low-carbohydrate/high-protein diets with low-fat/high-carbohydrate diets for obesity and cardiovascular disease risk. This systematic review focuses on randomized controlled trials of low-carbohydrate diets compared with low-fat/low-calorie diets. Studies conducted in adult populations with mean or median body mass index of > or =28 kg m(-2) were included. Thirteen electronic databases were searched and randomized controlled trials from January 2000 to March 2007 were evaluated. Trials were included if they lasted at least 6 months and assessed the weight-loss effects of low-carbohydrate diets against low-fat/low-calorie diets. For each study, data were abstracted and checked by two researchers prior to electronic data entry. The computer program Review Manager 4.2.2 was used for the data analysis. Thirteen articles met the inclusion criteria. There were significant differences between the groups for weight, high-density lipoprotein cholesterol, triacylglycerols and systolic blood pressure, favouring the low-carbohydrate diet. There was a higher attrition rate in the low-fat compared with the low-carbohydrate groups suggesting a patient preference for a low-carbohydrate/high-protein approach as opposed to the Public Health preference of a low-fat/high-carbohydrate diet. Evidence from this systematic review demonstrates that low-carbohydrate/high-protein diets are more effective at 6 months and are as effective, if not more, as low-fat diets in reducing weight and cardiovascular disease risk up to 1 year. More evidence and longer-term studies are needed to assess the long-term cardiovascular benefits from the weight loss achieved using these diets.

8 Evidence type: case studies

James R. Phelps, Susan V. Siemers & Rif S. El-Mallakh
The ketogenic diet for type II bipolar disorder.
Neurocase: The Neural Basis of Cognition
DOI: 10.1080/13554794.2012.690421

Successful mood stabilizing treatments reduce intracellular sodium in an activity-dependent manner. This can also be achieved with acidification of the blood, as is the case with the ketogenic diet. Two women with type II bipolar disorder were able to maintain ketosis for prolonged periods of time (2 and 3 years, respectively). Both experienced mood stabilization that exceeded that achieved with medication; experienced a significant subjective improvement that was distinctly related to ketosis; and tolerated the diet well. There were no significant adverse effects in either case. These cases demonstrate that the ketogenic diet is a potentially sustainable option for mood stabilization in type II bipolar illness. They also support the hypothesis that acidic plasma may stabilize mood, perhaps by reducing intracellular sodium and calcium.

Evidence type: review of hypotheses and controlled animal experiments

Carl E. Stafstrom and Jong M. Rho.
The Ketogenic Diet as a Treatment Paradigm for Diverse Neurological Disorders
Front Pharmacol. 2012; 3: 59.
Published online 2012 April 9. Prepublished online 2012 January 25. doi: 10.3389/fphar.2012.00059

Mood stabilizing properties of the KD have been hypothesized (El-Mallakh and Paskitti, 2001), but no clinical studies have been conducted as of this writing. The potential role of the KD in depression has been studied in the forced choice model of depression in rats, which led to a beneficial effect similar to that afforded by conventional antidepressants (Murphy et al., 2004; Murphy and Burnham, 2006).

9 Evidence type: review of human and animal experiments

Carl E. Stafstrom and Jong M. Rho.
The Ketogenic Diet as a Treatment Paradigm for Diverse Neurological Disorders
Front Pharmacol. 2012; 3: 59.
Published online 2012 April 9. Prepublished online 2012 January 25. doi: 10.3389/fphar.2012.00059

Clinical studies to date have been equivocal but promising. A randomized double-blind, placebo-controlled trial of a MCT KD resulted in significantly improved cognitive functioning in APOε4-negative patients with AD but not in patients with a APOε4 mutation (Henderson et al., 2009). In this study, the primary cognitive end-points measured were the mean change from baseline in the AD Assessment Scale-Cognitive subscale, and global scores in the AD Cooperative Study – Clinical Global Impression of Change (Henderson et al., 2009). This significant clinical improvement was considered to be secondary to improved mitochondrial function, since ketone bodies (specifically, beta-hydroxybutyrate or BHB) have been shown to protect against the toxic effects of β-amyloid on neurons in culture (Kashiwaya et al., 2000). Alternatively, the KD may actually decrease amounts of β-amyloid deposition (VanderAuwera et al., 2005).

[T]here is growing evidence that the KD may be an effective treatment for AD through a variety of metabolism-induced mechanisms that reduce oxidative stress and neuroinflammation, and enhance bioenergetic profiles – largely through enhanced mitochondrial functioning.

10 Evidence type: plausible mechanism and case studies

Seyfried TN, Marsh J, Shelton LM, Huysentruyt LC, Mukherjee P.
Is the restricted ketogenic diet a viable alternative to the standard of care for managing malignant brain cancer?
Epilepsy Res. 2012 Jul;100(3):310-26. doi: 10.1016/j.eplepsyres.2011.06.017. Epub 2011 Aug 31.


Malignant brain cancer persists as a major disease of morbidity and mortality. The failure to recognize brain cancer as a disease of energy metabolism has contributed in large part to the failure in management. As long as brain tumor cells have access to glucose and glutamine, the disease will progress. The current standard of care provides brain tumors with access to glucose and glutamine. The high fat low carbohydrate ketogenic diet (KD) will target glucose availability and possibly that of glutamine when administered in carefully restricted amounts to reduce total caloric intake and circulating levels of glucose. The restricted KD (RKD) targets major signaling pathways associated with glucose and glutamine metabolism including the IGF-1/PI3K/Akt/Hif pathway. The RKD is anti-angiogenic, anti-invasive, anti-inflammatory, and pro-apoptotic when evaluated in mice with malignant brain cancer. The therapeutic efficacy of the restricted KD can be enhanced when combined with drugs that also target glucose and glutamine. Therapeutic efficacy of the RKD was also seen against malignant gliomas in human case reports. Hence, the RKD can be an effective non-toxic therapeutic option to the current standard of care for inhibiting the growth and invasive properties of malignant brain cancer.