Ketogenic diet and STZ-induced diabetes

High fat diets cause diabetes. At least this is what we are told. Researchers frequently use streptozotocin (STZ) to induce diabetes in experimental animals. So, following the logic, a low carbohydrate ketogenic diet (LCKD) plus STZ would make rats extremely diabetic, with a very reduced chance to survive in the long term. 

So let's see what happens when STZ-treated rats are fed a normal chow diet (ND), a LCKD and a high carbohydrate diet (HCHO) (1). The macronutrient ratios for the latter were (C/F/P): LCKD 10/60/30 and HCHO 70/10/20.

Bodyweight remained constant in the LCKD group, while it was reduced significantly in the HCHO and ND groups. In the latter, after the administration of STZ, blood glucose (BG) increased from 105mg/dL at baseline to 650mg/dL at the end of the experimental period. In contrast, the LCKD group maintained BG levels around 100mg/dL. Food intake also was drastically increased in HCHO and ND groups, showing polyphagia. The LCKD rats showed a little increased in food intake, then decreased and remained constant during the whole study. Water intake was also constant in the LCKD compared to HCHO and ND. Urine output was also increased in the latter groups. (Remember the "three P's" of diabetes: polydipsia, polyphagia and polyurea). Glucosuria after STZ injection reached 1000mg/dL. However, LCKD showed negative glucosuria. Summing up: LCKD rats didnt show any marker of diabetes comared to HCHO and ND rats. They maintained calorie intake, weight and BG levels normal. No polydipsia, polyphagia or polyurea. 

One recent study warned about the mechanism by which high fat diets could cause diabetes and beta-cell dysfunction. Yes, this is the famous study by Ohtsubo et al (2). For a more comprehensive review of this study please refer to the one written by Denise Minger. In a nutshell, what the authors found was that elevated concentration of free fatty acids (FFA) caused nuclear exclusion and reduced expression of FOXA2 and HNF1A transcription factors in beta cells. This resulted in depletion of GnT-4a glycosylation and glucose transporter expression, leading to beta-cell dysfunction. This is one mechanism by which lipotoxicity contributes to diabetes onset. However, STZ causes cell death in pancreatic beta-cells through methylation, the release of free radicals or by the formation of nitric oxide. The mechanism found by Ohtsubo might be reversible. Beta-cell destruction might not. This is one of the most important problems with advanced diabetes, and might be involved in the evolution of type 2 into type 1 diabetes (3). Thus, studies using models of beta-cell destruction might be more relevant for understanding the basis of autoimmune or chronic uncontrolled diabetes.

This leads us to the most interesting part of the resent study. The authors assessed the histology of the Langerhans islets in the different rats by H&E staining. 

Copyright © 2010 Elsevier GmbH. All rights reserved.

(a) and (b) show the sections of the pancreas from control HCHO and ND rats. Circles show islets of Langerhans and arrows show vacuoles. (d) and (e) are from diabetic HCHO and ND rats, respectively. As can be seen, there is almost no islet left after STZ administration. On the other hand, diabetic LCKD rats showed no reduction of islets compared to LCKD controls ((c) and (f)). 

To further assess the efect of the different diets on beta-cell destruction, the authors used Gomori's Chrome Alum Haematoxylin-Phloxine stain. 

Copyright © 2010 Elsevier GmbH. All rights reserved.

Beta-cells are stained blue, alfa-cells are stained red and delta-cells are stained pink. (a), (b) and (c) are control ND, HCHO and LCKD; (d), (e) and (f) are diabetic ND, HCHO and LCKD, respectively. 

Overall, there was a clear protection against beta-cell destruction in the diabetic LCKD rats, compared to diabetic HCHO and ND rats. However, the number of beta-cells in control rats was not different between groups. 

How can a ketogenic diet can prevent the onset of diabetes induced by STZ and a high-fat diet cause diabetes? Isnt a ketogenic diet a high-fat diet? First, a high-fat diet is not necessarily a ketogenic diet. The term "high-fat diet" is used without a consensus in the literature, so a high sugar-high fat diet might be promoted as a high-fat diet (this is why is EXTREMELY important to read the methods). Second, lipotoxicity is a major cause of metabolic dysfunction. However, lipotoxicity doesnt implies a high-fat diet. It implies dysregulation of lipid metabolism. If anything, a ketogenic diet should restore a normal lipid metabolism. Third, there is a difference in comparing in vitro results with in vivo results. I have highlighted the importance of this distinction before. Finally, diabetes is a highly complex disease. I believe that the most serious cases have definitely an immune component, so there is targeted destruction of beta-cells. The authors speculated that the ketogenic diet prevented diabetes by the antioxidant effect of ketone bodies (because one of the cytotoxic effects of STZ in beta-cells is mediated by the increase in free radicals).

In conclusion, saying that high-fat ketogenic diets cause diabetes is as silly as saying that high carbohydrate diets cause diabetes. There is an extreme metabolic flexibility present in healthy humans, which can adapt to a wide range of macronutrient ratios. Food toxins, as stressed by other authors are another source of problems, which can confound the effect of different diets. 


People with diabetes might benefit from low carbohydrate diets not only by the proximate effect of the diet (less dietary glucose, which treats the symptom, not the cause), but because of calorie restriction, which alleviates lipotoxicity. This can also be achieved with an hypocaloric high-carbohydrate diet. But with people in with autoimmune type I diabetes, LADA, or severe cases of type 2 diabetes, a ketogenic diet could prevent progression of the disease more efficiently, preventing oxidative stress-mediated cell death. 

Al-Khalifa A, Mathew TC, Al-Zaid NS, Mathew E, & Dashti H (2011). Low carbohydrate ketogenic diet prevents the induction of diabetes using streptozotocin in rats. Experimental and toxicologic pathology : official journal of the Gesellschaft fur Toxikologische Pathologie, 63 (7-8), 663-9 PMID: 21943927

Total Ketogenic Ratio (TKR)

The degree of ketosis produced by a particular diet has been long determined by the Ketogenic Ratio (KR), proposed by Woodyatt (1):

KR = K / AK

where, K represents the ketogenic factors and AK represents the anti-ketogenic factors. 

Experimentally determined values for each dietary component are expressed by the equation:

Where:

f = dietary fats in grams
p = dietary protein in grams
c = dietary carbohydrate in grams

Another easier way to determine the KR was proposed by Metcalf and Moriarty:

KR = f / p + c

The formula doesn't take into account the antiketogenic properties of some amino acids but gives an approximate result to the Woodyat equation. 

A paper published by I.A. Cohen (2) adressed the issue brought by these equations: they only apply to isocaloric conditions because they fail to take into account the utilization of energy stores in a hypocaloric diet, underestimating the degree of ketosis of a given diet. The Woodyatt equation then is referred as the Dietary Ketogenic Ratio (DKR), representing the effect of the dietary components on ketosis. 

The formula proposed by Cohen tries to model the impact of weight reduction dieting under ideal conditions, that is, the utilization of stored lipid without the catabolism of protein, scenario normally reached after a few days of fasting and/or low carbohydrate-high fat dieting. 

The Cohen formula is derived from the Woodyatt equation, in a series of simple, mathematical steps giving the new equation:

Where

e = total energy expenditure in calories
p = dietary protein in grams
c = dietary carbohydrate in grams

There are some limitations of this equation, as stated by the author:


1. It needs further evaluation for proving the hypothesis.
2. It describes an ideal situation in which any energy deficit is met entirely by stored lipids. 
3. Dietary energy established by bomb calorimetry may exceed from the energy obtainable by the organism consuming that diet.
4. Energy efficiencies brought about by dieting, as well as changes in the use of dietary nutrients for non-energy uses may alter the expected outcome. For instance, subjects who have not utilized particular metabolic pathways for an extended period may show changes in energy utilization over time.
5. Urinary loss of ketones is not accounted for in these formulae.
6. The nutrient mix itself, as well as the state of the organism may influence the ability to fully utilize the energy available in the consumed nutrients.
7. There is inter-individual variability for achieving the same degree of ketosis. Genetic factors and differences in intestinal flora may alter this parameter.
8. Woodyatt analysis of the ketogenic and anti-ketogenic properties of protein depends upon the proportions of specific amino acids, raising the possibility of differing properties depending upon protein sources. This may necessitate having to alter Woodyatt's constants depending upon protein mix.
9. Differences on both the absorption and utilization of fats resulting from the types of fat included in the diet might alter the response, ie. saturated vs. unsaturated, polyunsaturated vs. monounsaturated, inclusion of MCT or the presence of abnormalities in individual gall bladder functioning.
10. The antiketogenic effects of carbohydrates in food might be altered by the rapidity of their conversion to glucose, as expressed in their glycemic index. The timing of meals may alter the results, particularly if individual feedings varied greatly in their ketogenic to anti-ketogenic properties. For example, meals that contained an isolated bolus of carbohydrate might be expected to rapidly decrease ketosis, even when the total amount of carbohydrate would not be expected to do so when evaluated over a 24-h period. 
11. Ketosis is but one factor among many in weight-loss dieting. The behaviour of any particular factor, by itself, should not imply suitability nor superiority. 


Let's see some examples. 


Asume that an hypocaloric diet consisting of 100g of protein, 70g of fat and 20g of carbohydrate, for an energy expenditure of 2000kcal:


KR (Woodyatt) = 1.28
TKR (Cohen) = 2.09

The energy supplied by the diet in this example is 1110 kcal, which creates a 900kcal deficit. This additional fat increases the production of ketones, not accounted by the Woodyatt equation.

If we up the protein intake to 150g while maintaining calories, we get:

KR (Woodyatt) = 1.00

TKR (Cohen) = 1.65

So even if we consider a PSMF (Protein Sparing Modified Fast) type diet, 200g of protein with 20g of fat and 20g of carbohydrate, we get:

KR (Woodyatt) = 0.79

TKR (Cohen) = 1.37


While the Woodyatt formula might underestimate the level of ketosis, it is probable that the Cohen formula overestimates it, specially in obese and overweight subjects. 

Cohen IA (2009). A model for determining total ketogenic ratio (TKR) for evaluating the ketogenic property of a weight-reduction diet. Medical hypotheses, 73 (3), 377-81 PMID: 19410378

Beta hydroxybutyrate might make you smarter

During physiological ketosis, KB can supply almost 60% of the brain's energy requirement. Plasma KB pass the BBB through monocarboxylic acid transporters (MCT) in a gradient-dependent manner and are readily available for neurons and astrocytes (which are also ketogenic (1)). In fact, the brain is happy without glucose and using ketones (2). Besides being neuroprotective and metabolically more efficient, the main KB, bOHB (3-HB), could enhance memory and learning.

Zou et al (3) tested this hypothesis in mice:

"This study sought to investigate the effect of 3-HB and derivatives on neuroglial cell metabolic activity and gap junctional intercellular communication of hippocampal neurons, to evaluate the hippocampal expression of PUMA-G and proteins related to memory following treatment with 3-HB, and to determine whether 3-hydroxybutyrate methyl ester (3-HBME) improves learning and memory in the normal mouse."

Interneuronal communication can occur indirectly and directly. In the former, chemical synapses are involved, in which transmitters are released into the extracellular space and bind the postsynaptic cell membrane. In the latter, electrical synapses mediate communication. The most prevalent group of electrical synapses, neuronal gap junctions, connect directly the intracellular space of two cells by gap junction channels. Connexin proteins are the structural components of gap junction channels in the nervous system. Specifically, connexin 36 has been involved in learning and memory (4).

3-HB is the endogenous ligand for PUMA-G (5). Its activation in adipose tissue is also produced by nicotinic acid, inhibiting lipolysis and controlling the rate of ketogenesis. This represents an homeostatic mechanism by which 3-HB controls its own production, preventing ketoacidosis. 

Researchers used the Morris water maze for measuring learning and memory, and analyzed hippocampal neuron exposure to 3-HB and derivatives in vitro. All of the metabolites stimulated metabolic activity in neuroglial cells. 3-HMBE increased gap junction intercellular communication, as well as connexin 36 expression by 30% (compared to 12% in mice treated with acetyl-L-carnitine) and pERK2 levels (phosphorylated ERK2, necessary for connexin assembly). Moreover, PUMA-G mRNA was found in the hippocampus, subthalamic nucleus, temporal cortex and frontal cortex, and 3-HBME enhanced the transcription of PUMA-G in the hippocampus.

So this translates to improved memory and learning?

"The escape latency of mice in all groups decreased with time. Overall treatment comparisons indicated that statistically significant differences existed among the groups. Notably, the 30 mM 3-HBME groups took less time (p < 0.05) than the control groups on days 1 and 3–5 to find the platform (Fig. 5B). All treatment groups were faster (p <0.05) than the control group on day 5."

Treatment groups include mice treated with either 3-HBME (20, 30 or 40mg/kg/d) or acetyl-L-carnitine. Control group was water. 

"Paths taken to the platform area on the fifth day of spatial training by mice in the 30 mg/kg/d 3-HBME group were more direct than those taken by mice in the control group, which took more circuitous paths."

Ketotic mice knew exactly what they wanted. 

This causes less total swimming distance (ie. increased efficiency):

"Similar to the escape latency results, the total swimming distance of the 30 mM 3-HBME groups was shorter (p <0.05) than the control groups on days 1 and 3–5. (...) Moreover, the total swimming distance of mice in the 30 mg/kg/d 3-HBME group was shorter (p <0.05) than that of the control or other treatment groups at days 3–5."

A probe test* then was performed to evaluate memory. Three parameteres were measured: the number of crossings of the exact place where the platform had been located, the swimming distance in the quadrant of the former platform position, and the swimming path in the pool. 

Number of times that mice crossed the former position of the hidden platform within 60s:

 

Swimming distance in the platform quadrant:

Swimming path:

Finally, thigmotaxis was lowest in the 30mg/kg/d 3-HMBE group.

A retention test was performed two days after the probe test: 

These test showed that the 30mg/kg/d 3-HMBE group:

a. Crossed more times the exact place where the platform had been located,
b. Had the larger swimming distance in such quadrant,
c. Found the platform faster than the other groups, 
d. Were the most calmed (less thigmotaxis), and
e. Found the platform faster than the other groups during the retention test

These results suggest that 3-HMBE, given at 30mg/kg/d, enhanced learning and memory**.

Researchers used polyhydroxybutyrate (PHB) (a polyhydroxyalkanoate) to produce 3-HB and derivatives. Can we extrapolate these results to physiologically produced 3-HB? The finding that hippocampal neurons expressed PUMA-G receptors is encouraging. As mentioned, 3-HB is the endogenous ligand for PUMA-G, but there seems to be a desensitization of the receptor when exposed to large amounts of 3-HB compared to 3-HBME and 3-HBEE. As PUMA-G interaction with several ligands is not well understood, we cannot draw many conclusions. 

Some studies have evaluated the potential benefit of 3-HB in neurological disorders which compromise memory and learning, such as Alzheimer's disease (AD). For example, in some patients with AD and mild cognitive impairment, MCT oil produces an improvement on cognitive testing and paragraph recall (6), correlated with the increase in plasma ketones. This neuroprotective effect has been explained by the increase in metabolic efficiency associated with 3-HB (7) and inhibition of apoptosis (8), but it might also act through PUMA-G and connexin 36 dependent mechanisms. Further studies should help discovering the molecular pathways involved. In the meantime, maybe schools should start giving students coconut oil shots instead of skim milk.

* The platform was removed from the pool and mice were challenged to a single search trial for 60s. 
** Note that this group performed best than 20mg/kg/d and 40mg/kg/d.


Zou XH, Li HM, Wang S, Leski M, Yao YC, Yang XD, Huang QJ, & Chen GQ (2009). The effect of 3-hydroxybutyrate methyl ester on learning and memory in mice. Biomaterials, 30 (8), 1532-41 PMID: 19111894

Ketones fuel fetal development

Ketosis during pregnancy has been known for many years. Fetal growth depends on constant energy supply, so physiological mechanisms should have been developed during evolution to assure intra-uterine development under starvation or food scarcity. Most studies focusing on pregnancy and fetal development have been done, for obvious reasons, on animals. It is not possible to extrapolate every detail, but it gives us a great idea and explanation for the metabolic changes observed during pregnancy. 

Briefly, there are two metabolic periods clearly differentiated during gestation. The first one, corresponding to the first two thirds, is the anabolic phase characterized by hyperphagia and enhanced storage of body fat (we will call it Phase I). During the last third of gestation, the catabolic phase, fetal growth is very rapid, so the energy needs of the fetus are increased (1) (we will call it Phase II). Insulin metabolism, as an acquired evolutionary mechanism, plays a key role during this process. During Phase I, there is a 3.0 to 3.5 fold increase in first-phase and second-phase insulin release in response to glucose, without an alteration in peripheral IS (2). This assures that accumulation of protein, glucose and fat is appropriate for late pregnancy.  As pregnancy progresses, this increase in glucose-stimulated insulin secretion is maintained, but IS is reduced in 50-70% (3, 4) during late pregnancy (Phase II). This mechanism serves to redistribute glucose and energy to the rapid growing fetus. In addition to peripheral IR (but not hepatic), gluconeogenesis (GnG) is increased 16 to 30% to supply the placenta and fetus demand. Contrary to the main GnG precursors in non-pregnant adults, glycerol is the main glucose precursor, which represents a mechanism by which in the abscence of food, the mother is capable of producing the necessary glucose from a substrate that is readily available during fasting and not depend on external substrates. This process is accentuated by fasting, commonly known as "accelerated starvation": compared to non-pregnant, women during gestation exhibit a pronounced hypoglycemia and rapid rise in KB. GnG increases parallels the rise in KB (4). Because of its increased utilization, glucose has drawn much attention away from the importance of KB in fetal development. 

bOHB is utilized in a dose-dependent manner by the rat conceptus (5) and serves to spare glucose and lactate for biosynthetic pathways (6). bOHB seems to be the main oxidative fuel to the human fetal brain, measured by the production of CO2 (7). A classic study done on rat embryos underscore the importance of both glucose and bOHB to a proper development (8). Researchers tested the effect of increasing doses of glucose, KB or both on organ teratogenesis. They first tested glucose alone. According to the authors:

(...) we found that isosmotic supplementation of the culture medium with 12 mg/mL D-glucose during the 48-h incubations effected a generalized retardation of rat-embryo growth and lesions such as microencephaly, exencephaly, open neural tube, and pericardial edema (6). We documented specificity by demonstrating that the findings are not replicated with isosmotic equimolar additions of certain other hexoses, such as sorbitol, fructose, inositol, or galactose (6). Teratogenic potentialities of high glucose concentrations have also been demonstrated with cultured mouse embryos. Sadler elicited dysmorphogenic effects with increasing frequency by adding 5mg/mL or 8 mg/mL D-glucose to the suspending rat serum during mouse-embryo culture (33).

So high glucose concentrations are teratogenic for the embryo. They went further and examinated the effect of increasing doses. 

During the period of these studies in 1980-1981, isosmotic additions of 12 mg/ mL elicited a 49% incidence of minor and a 23% incidence of major lesions. By contrast isosmotic additions of 3 mg/mL D-glucose to the incubation medium did not evoke any discernible lesions during 48 h of culture, 6mg/mL resulted in only a 2.2% incidence of minor and no major lesions, and 9 mg/mL D glucose were required to elicit 5.1% major and 17.8% minor lesions in the cultured intact embryos from our outbred strain of Charles River Sprague-Dawley rats.

They concluded:

(...) the dysmorphogenic potentialities of ambient glucose are clearly concentration dependent although the precise relationships may be quantitatively different in various species or in different strains from the same species.

So we know that hyperglicemia is teratogenic. But what about increasing doses of bOHB? 

Preliminary acute incubations with 14C-labelled 14C-hydroxybutyrate indicated that cultured embryo units can oxidize ketones on day 10.4 as well as 1 1.4 of development (36) so that ketones can subserve nutrient functions in some portions of the conceptus at both times. What about the effects of ketones on embryogenesis during these intervals? As summarized in Figure 3, isosmotic additions of 2 or 4 mM buffered D,L sodium (3-hydroxybutyrate during 48-h culture of rat conceptus from day 9.5 to 1 1 .5 of development did not elicit any discernible dysmorphogenesis.

So physiological concentrations of bOHB, as in a low carbohydrate diet, ARE NOT TERATOGENIC. Problems appear only when going above this threshold, as in DK. 

However, with 8 mM, 24.5% of the embryos developed minor lesions, and the inclusion of 16 mM D,L /3-hydroxybutyrate was associated with a 71% frequency of minor and 45% incidence of major lesions (36). 

See the trend? With 8mM only a quarter developed minor lesions. But when levels went way up (not physiological) lesions are aggraviated.

On the left, added concentrations of glucose and on the right, added concentrations of bOHB. The trend is clear, there is no damage when KB are in the physiological range, but when levels increase to concentrations seen in DK, boom! As always, the problem arises with hyperketonemia, not ketosis. Its easier to develop hyperglycemia than hyperketonemia (except during starvation).

Lastly, what happens if we mix the minimally teratogenic amount of glucose (6mg/dL) with the minimally teratogenic amount of bOHB (8mM)? Sinergy! 66% displayed minor lesions and 27.7% major lesions. Some of the effects could not be explained by normal growth retardation. 

KB are so important to normal growth that there is evidence that fetal ketogenesis occurs (9). To achieve an optimal development, the fetus must not be exposed to increased concentrations of KB nor glucose. Both sources of fuel are necessary but in the right amount. The body adapts to this situation increasing the production of glucose from glycerol, reducing the need for ingesting extra glucose. Increasing calories and carbohydrates during pregnancy predisposes the mother to hyperglycemia, GD and IR, neonatal macrosomy and teratogenesis. Reducing the GL of the diet has shown to offer benefits compared to a low-fat diet (10), even when carbohydrate intake is reduced to 40-45% of total calories (11). Controlled studies adressing the effects of less than 40% of carbohydrates are scarce (evil ketosis!). Nevertheless, going zero carb can be as dangerous as going high carb* (12). But there is no need to go up to 60%. In rats, the requirement for normal growth seems to be around 18-20% (13), comparable amount of carbohydrates eaten by most low carbers and/or paleo, while the human fetus consumes around 20-25g/glucose per day during late gestation (4). 

Maintaining a proper diet with plenty of saturated fat, low carbohydrate and adequate protein/EPA+DHA is essential for a healthy pregnancy. Quality over quantity.

Any experiences to share?

*Although this is physiolgically impossible. Only achievable eating zero carb protein drinks and oil.

High protein and ketosis

Traditional ketogenic diets are both low in protein and carbohydrates. The main argument is that because almost 60% of the ingested amino acids are glucogenic, one must also restrict protein intake to achieve ketosis. According to VanItallie and Nufert (see Ketosis Essentials):

"This occurs because approximately 48 to 58% of the amino acids in most dietary proteins are glucogenic. For every 2 grams of protein consumed in a carbohydrate-free diet, somewhere between 1.0 and 1.2 grams are potentially convertible to glucose. Therefore, to obtain a degree of hyperketonemia (approximately 2–7 mM/L) believed to be therapeutically effective in certain important medical conditions such as epilepsy, patients must rigorously restrict protein as well as carbohydrate intake and, when possible, increase their level of physical activity. (my emphasis)"

This is why, for example, a typical ketogenic diet to treat epilepsy utilizes a ratio of 4:1 or 3:1 (Fat:Protein+Carbohydrates). 

So what happens when we eat a high fat/high protein diet? Let's look at Eskimos (1). On average, they consume approximately 280g of protein, 125g of fat and 54g of carbohydrate. This gives us a total of 2461 kcal of which 45.5% is protein, 45.71% is fat, and around 8.77% is carbohydrate. This is both high protein and high fat, and very low carbohydrate. Kind of my ideal diet. When researchers studied their metabolisms, they found that they are not in ketosis during their usual diet. Ketosis is developed during fasting, but to a much lesser degree than other human subjects. The authors concluded:

"Eskimos show a remarkable power to oxidize fats completely, as evidenced by the small amount of acetone bodies excreted in the urine in fasting." 

Further studies showed the same results. For instance, Steffanson and Andersen, both who lived eating 9 years an Eskimo diet, participated in a controlled study eating only meat for one year (2). Besides health improvements, a very mild ketosis was observed, similar to the Eskimo studies. But these studies are really old (late 20's-early 30's) and the degree of ketosis was measured by urinary ketones. It is known that ketonemia is a better indicator of the degree of ketosis than ketonuria (3, 4, 5). This is very important because measuring ketosis by ketonuria tend to show many false positives (throw away dose damn strips!). 

During popular weight loss ketogenic diets, ketonuria (6, 7) and ketonemia (8) are observed despite the high percentage of calories derived from dietary protein. Even with a low calorie high protein diet (PSMF) ketonemia appears (9).

If KB are excreted by urine, why does ketonuria not always correlate with the degree of ketosis? The answer lies in KB metabolism. During fasting, KB start to rise until a plateau is reached almost at 5 days. This occurs both because of a reduced skeletal muscle clearance and decreased production by a negative feedback loop (10). The effect on KB removal rate is equal on both diabetic and normal subjects, only differing on the production rate of KB (ketogenesis)*. Urinary excretion of KB is always < 10% of total turnover. So the level of KB in plasma is determined by the difference between ketogenesis and clearance by extra hepatic tissues, while ketonuria accounts only for a small part of the equation. Increasing the utilization of ketones by peripheral tissues reduces its excretion, so one active keto-adapted person can be in strong ketosis but show almost no urinary ketones.

Although the degree of ketosis is dependent on the amount of dietary fat, there are some tools for increasing it without decreasing protein intake and/or increasing fat intake. Exercise has shown to increase both ketogenesis and metabolic clearance rate (11), but the effect on the latter is abolished at high concentrations (12) and when basal ketone concentrations are high (5.7mM) (13). This is because skeletal muscle adapts to use FFA as energy and spares ketones for non-FFA-using tissues, like the brain. During a LCKD, exercise enhances ketogenesis and the degree of ketonemia, as RQ values during exercise in ketogenic conditions have shown to be as low as 0.7 (14) and even 0.66 (15). In fact, post-exercise ketosis is a well known phenomenon (16). 

The hormonal environment is another factor that influences ketogenesis and plays a direct role in the enhancing properties of exercise. Insulin is the classic anti-ketogenic hormone, while catecholamines are strong ketogenic activators (17, 18). Exercise increases the body energy needs and catecholamine secretion, while decreasing insulin and glucose levels (16). With some differences, this scenario is similar during fasting. Traditionally, studies on FK and energy metabolism during fasting have been done during starvation, but lately ADF and IF is being studied as a potential therapy to treat and prevent several diseases. For example, ADF increases bOHB levels after 22 days even while eating ad libitum and without  carbohydrate restriction (19). Moreover, KB begin to rise after an overnight fast in "normal" people**. This is because during short term fasting the expression of PDK4, LDL, UCP3 and CPTI is increased, showing a "glucose-sparing" mechanism and shifting towards a lipolytic metabolism (20). This changes arent reversed by a LCKD, which has similar effects on gene expression. IF/ADF would potentiate the effects of a LCKD on ketogenesis, lipid metabolism and glucose sparing.

Another aspect which influences the degree of ketosis is food. It is well known that coconut oil (and specifically MCTs) increases plasma KB (21, 22). Black tea has also shown some ketogenic properties (23). Adrenergic stimulation by dietary stimulants (such as coffee or green tea) could induce ketogenesis, specially during a fast, where adrenergic sensitivity is increased. Both adrenaline and noradrenaline have ketotic effects (24). Catechin-poylphenols in green tea, for example, have shown to inhibit catechol-O-methyl-transferase and caffeine to inhibit transcellular phosphodiesterases (25).

Summary

You can eat a high protein diet and still induce strong ketosis. A combination of a high fat/high protein diet, resistance exercise and fasting (as per my recommendations in my introduction post) induces a strong metabolic response of adaptation to a ketotic environment, without worrying on specific macros or restricting too much protein intake.

Addendum: That certain amino acids are glucogenic means that they can potentially be converted to glucose. In this case, when keto-adapted, glucose needs are reduced, so is GnG. This has been observed in studies in which a severe carbohydrate restriction only increases GnG slightly after a few days (26, 27). Increasing glucogenic precursors per se does not rise BG levels (28)***, which might affect ketosis. Hepatic glucose output is an extremely well regulated process which serves to mantain BG levels in an adequate range and prevent hypoglycemia.  As keto-adaptation occurs, glycogenolisis is reduced and GnG increased, in a controlled manner. GnG must not be avoided, its the natural and optimal way to control glycemia****.


*The reason why ketoacidosis is not developed under non-diabetic circumstances is because of insulin, which controls the rate of ketogenesis and lipolysis. 
** "Normal" as someone who eats SAD or a high carbohydrate diet.
*** This is why protein does not rise BG levels.
**** In TIIDM subjects, for example, this HGO control is dysregulated so hyperglycemia occurs.