Thursday, January 12, 2012

Safe starches, blood glucose and insulin

A reader asked me recently about a subject which is confusing many people in the paleosphere. 
"Jaminet in his debate with Rosedale suggests higher carb diets tend to lower blood sugar whereas low carb diets, RAISE it. Can you help me untangle what is going on here? It makes it sound like the more carbs you eat the better your blood sugar levels which does not seem right to me. Clearly, an important health goal is achieving low blood glucose so I would want to know what is the best way to eat to control them. 
I always assumed that the less sugar you eat, the less blood sugar you'll have. Is there a threshold? If Jaminet is correct, then shouldn't we see a higher fasting glucose associated with ketogenic diets that are totally carb restricted vs higher carb diets? I know excess protein might be converted into glucose but if you followed a ketogenic diet with low protein would you still see the rise in blood glucose? 
What is the mechanism through which blood glucose is being lowered in high carbers? Do they secrete more insulin to deal with it hence lower blood glucose? Would their insulin levels therefore be higher even if blood glucose was relatively low? Which is worse for health- low glucose/high insulin or moderate glucose/low insulin? 
What is the mechanism through which some cancers are being suppressed with ketogenic diets if not through lowered blood glucose? Is there something else going on?"
These are very important questions as they raise some concern in people which utilize a low carbohydrate diet for controlling their blood glucose (BG). Before trying to elaborate an answer, there are some facts that must be kept in mind:

  • Hyperglycemia is not a disease, it is a symptom. 
  • BG levels can be affected by non-dietary factors.
  • The two principal energy substrates for humans (glucose and free fatty acids (FFA)) compete with each other for their utilization.
  • Calories matter.
  • There are differences between physiological insulin resistance (PIR) and pathological insulin resistance (PaIR). The term "insulin resistance" is very vague, it doesn't define explicitly which tissue(s) is IR.

The first issue to adress is whether ketogenic diets raise BG levels. The only evidence I have seen for this happening is in anecdotes from people in the internet. But if we want to have an objective look at the subect, we must see what happens in studies done with ketogenic diets (I will use low carbohydrate and ketogenic diets equally). 


Ketogenic diets and blood glucose levels

Most studies done on ketogenic diets are short-term and involve weight loss. All of them show a reduction in BG and insulin levels. A study by Grieb et al. (1) found that people eating an optimal diet (Kwasniewski) had on average a BG level of 87.9mg/dL, which is in the normal range; and very low values of HOMA-IR. Sharman et al. (2) have shown that the metabolic benefits of carbohydrate restriction are independent of weight loss. Given the evidence, it is only possible to speculate about the mechanisms by which BG levels rise in some people eating a ketogenic diet.

The goal of a ketogenic diet is to simulate fasting, but without the negative effects of prolonged nutrient restriction. Before going on, it is pertinent to remember the Randle Cycle (3):


In short, both FFA and glucose compete with each other for their uptake and oxidation. This can be translated as when BG is high, FFA utilization is low; and when BG is low, FFA utilization is high. Ketogenic diets are characterized by low BG levels, in part because of a drastic reduction in exongeous glucose. In turn, plasma FFA rise from dietary and endogenous sources (the contribution of each one depends on energy balance). During this scenario, plasma ketone bodies also rise. So we have the following metabolic milieu:


Cellular effects of a fasting-type metabolism

For exerting it metabolic effects, insulin needs to first bind its receptor (the insulin receptor, IR). Upon binding, insulin triggers an intracellular signaling cascade, which influences both the function of intracellular proteins and gene expression. The signaling pathway triggered by the binding of insulin include the recruitment of the IRS (insulin receptor substrate) to the cytosolic part of the insulin receptor dimer. The signaling cascade stimulated by insulin is essential for its function. If proteins involved in this signaling pathway are inhibited, there will be no cellular response to the binding of insulin and the activation of IR.

One level of inhibition of glucose utilization by FFA involves the inhibition of GLUT4 translocation to the plasma membrane. The translocation of GLUT4 is stimulated by insulin, by activation of IRS, PI3K and other proteins (4, 5). In vitro studies have shown that palmitate, the main fatty acid stored in mammalian adipose tissue, inhibits GLUT4 translocation and activity (6, 7). This results in reduced glucose uptake in skeletal muscle.

Inactivation of PDH (pyruvate dehydrogenase) is one of the most important mechanisms for inhibition of glucose oxidation  by FFA. PDH activity is controlled by phosphorylation, by PDK (pyruvate dehydrogenase kinase) and PDP (pyruvate dehydrogenase phosphatase). Phosphorylation by PDK inactivates PDH, while dephosphorylation by PDP activates it. Fatty acid oxidation increases the mitochondrial ratios of [acetyl-CoA]/[CoA] and [NADH]/[NAD+], which inhibit PDH. Low carbohydrate diets have shown to reduce muscle PDH and increase PDK (8, 9), effects which are reversed by a carbohydrate refeeding (10). Fatty acids also increase the concentration of cytosolic citrate, which inhibits 6-phosphofructo-1-kinase, providing another mechanism of inhibition of glucose oxidation (3). Fatty acids can reduce phosphorylation of IRS-1 (11), GSK-3b and PKB/Akt (12), thereby acting also downstream of IR and IRS.

The metabolic response to fasting can show what are the effects on insulin signaling of very high levels of plasma FFA and ketone bodies, but within a physiological range. In a very interesting study, Soeters et al (13) found that insulin-mediated peripheral glucose uptake after 62h of fasting was significantly lower compared to 14h of fasting. They also found that after 62h of fasting, Akt phosphorylation at Ser473 and AS160 phosphorylation at Thr642 were reduced. This implies that insulin signaling was attenuated (reduced phosphorylation of Akt) as well as glucose uptake (phosphorylation of AS160 is involved in the translocation of GLUT4). The authors concluded:
"(...) it is possible that pAKT-ser473 is involved in the physiological adaptation to fasting, inducing a reduction in peripheral glucose uptake and protecting the body from hypoglycemia."
Intramyocellular triglyceride accumulation is thought to mediate fatty acid insulin resistance. This is one way by which some authors think that a high-fat diet leads to insulin resistance. Compared to fasting (67h) a very low carbohydrate diet (eucaloric) produces the same amount of IMTG accumulation, both produce glucose intolerance and reductions in insulin sensitivity (14). Thus, the factor for triggering this metabolic response seems to be the absence (or drastic reduction) in glucose availability (my bolds):
"Thus, we suggest that dietary-induced IMTG accumulation and insulin resistance in healthy humans may be largely influenced by circulating FFAs, whose availability (in turn) is regulated by dietary CHO intake. (...) our study provides support for the hypothesis that the physiological trigger for this coupling in the healthy individual may be a short-term challenge to dietary CHO availability. That we have observed these diabetogenic alterations in a physically fit population, which is purported to be insulin sensitive yet exhibits high IMTG concentrations (the ‘athlete paradox’) (Goodpaster et al. 2001), supports our contention that they represent an adaptive rather than pathological response. This substantiates our previous assertion that alterations in glucose tolerance and insulin sensitivity associated with dynamic changes to the plasma and/or lean tissue lipid profile are part of a normal co-ordinated adaptation to short-term changes in food availability (Stannard & Johnson, 2004) and perhaps, more specifically, to fluctuations of dietary CHO availability. (...) This short-term alteration is teleologically sound because it limits competition between skeletal muscle and glucose obligate tissues for circulating glucose substrate when its availability becomes limited. Irrespective of a causal relationship, the coupling between IMTG accumulation and reduced insulin sensitivity may also represent a co-ordinated adaptive (non-pathological) response to CHO stress  (Johnson et al. 2003). A concomitant resistance in muscle to the effects of insulin on glucose uptake during CHO stress maintains normoglycaemia and thus the preservation of plasma glucose for use by the CNS and glucose-obligate tissues (Reaven, 1998). Dissociation of insulin action by way of muscle insulin resistance rather than attenuation of insulin secretion means that residual circulating insulin levels can be maintained (Klein et al. 1993), thereby preventing rampant proteolysis (Fryburg et al. 1990), lipolysis (Kather et al. 1985) and perhaps hepatic glucose release, whilst unnecessary uptake of blood glucose by muscle is prevented."
So, from the studies above, we can conclude that:

  • FFA supress glucose uptake and oxidation, resulting in muscular insulin resistance, without reducing insulin secretion.
  • These effects seem to be dependent on dietary carbohydrate restriction.
  • FFA-induced muscular insulin resistance is a physiological response to low availability of glucose. 
  • Under normal conditions, this serves to maintain adequate BG levels. When FFA are in excess, there might be a rise in BG levels, because oxidation and release of FFA are not coupled. This leads to insulin resistance in other tissues like the liver (15), consequently failing to control hepatic glucose output.

Blood glucose levels and high carbohydrate diets

Without any biochemical explanation, logic dictates that if we eat a high carbohydrate diet, glucose oxidation pathways are stimulated. This is the opposite of what we observe with carbohydrate restriction, that is, stimulation of insulin signaling. Glucose and insulin both regulate GLUT4 and GLUT1 in muscle cells (16) to increase glucose uptake. The Randle cycle dictates that glucose stimulates its own oxidation and reduces fatty acid utilization. As shown above, glucose reduces PDK and increases PDH. Insulin inhibits lipolysis, further facilitating glucose oxidation. In healthy subjects, a high carbohydrate-low fat diet can improve insulin sensitivity (17, 18). It makes sense, carbohydrates supress fat oxidation and increse glucose oxidation. Overall, there should not be a rise in BG levels on a 24h basis, if anything, we can expect a reduction, because we are utilizing glucose as our main substrate. 

The common ground: calorie restriction

Until now, we have seen that carbohydrates stimulate glucose utilization (insulin sensitivity) and that FFA supress it. How can then a ketogenic diet produce such good results in people with diabetes? Diabetes and MetSyn are characterized by lipotoxicity and glucotoxicity (19). This means that there is an abnormal level of plasma FFA and glucose, produced by PaIR. Insulin cant supress hepatic glucose output, muscle cells do not respond to insulin, and adipocytes liberate FFA in an uncontrolled fashion. In very simple terms, there is an excess of both energy substrates, each one inhibiting the utilization of the other. This scenario can be improved both by restricting fat (thereby increasing glucose utilization) or restricting carbohydrates (increasing fat utilization). In either case, calories must be restricted (directly or indirectly). So, people who show signs of glucose intolerance and switch to a ketogenic diet can improve their BG and insulin levels (by reducing glucotoxicity), but if energy is in excess, BG can start to rise. On the contrary, reducing dietary fat alleviates lipotoxicity, increasing insulin sensitivity. This is why any diet which is calorie restricted, independent of macronutrient composition, produces weight loss and improves glucoregulation. Calorie restriction, by producing a calorie deficit, alleviates both gluco- and lipotoxicity. 

One of the important aspects for dealing with this subject is the fact that glucose intolerance can have many underlying causes. In this manner, a person with autoimmune diabetes may not tolerate carbohydrates as well as a person with only mild PaIR. The fact that PaIR may progress into beta-cell dysfunction (20, 21) can alterate the response to a high carbohydrate diet, and depending on the severity, extreme measures must be taken to achieve normal BG and insulin levels (such as severe calorie restriction). The distribution of body fat can also have consequences on glucoregulation (22). Last but not least, epigenetic changes produced in utero can affect glucose tolerance since the moment we are born (23). 

What is more important, in my opinion, is to address whether hyperinsulinemia causes or potentiates IR, or if hyperinsulinemia results from IR, by a compensatory mechanism. If the first hypothesis holds true, then a ketogenic diet would have an advantage over a low fat-high carbohydrate diet. Desensitization of target cells triggered by the same hormone (homologous desensitization) is a very common characteristic of hormone signaling. In short, high levels of a given hormone reduce the response of the cell to the hormone effects and a reduction in the level of this hormone resets sensitivity. Excess hormone signaling is harmful, so the cell's attempt to restore normality is mediated by reducing its response. This is exactly what happens with insulin: 
  • Chronic hyperinsulinemia (in vivo and in vitro) causes a reduction in the number of receptors per cell and glucose transport (24, 25).
  • Pre-incubation of 3T3-L1 adipocytes with high levels of insulin and glucose increase PTEN activity, which is correlated with decreased PtdIns(3,4,5)P3 (26). This metabolite is very important for intracellular signaling transduction of insulin.
  • Hyperinsulinemia has shown to induce insulin resistance in humans (27).
  • Overall, hyperinsulinemia is proposed to be a result and a driver of insulin resistance (28).
Obesity seems to be characterized by an increased amount of insulin being secreted, compared to lean subjects (29). So, while calorie restriction per se is responsible for improved glucoregulation, there might be a short-term benefit in consuming a high-fat ketogenic diet in T2DM and MetSyn patients. As the bodyfat mass and associated hormones regulate, the differences between hypocaloric diets with different macronutrient profiles might be eliminated. This seems reasonable for diet-induced insulin resistance, but not for autoimmune or severe diet-induced glucose intolerance. There seems to be a threshold in which many people cant fully recover their insulin sensitivity with dietary measures. This is where a more integrative and previously uncharacterized approach kicks in (this is the subject of my future post). 

Glucose and cancer

Glucose restriction for cancer treatment seems reasonable given the evidence on the dependence of most types of cancer on glucose for cell growth and proliferation. Unfortunately, the picture is not that simple (30). Although restricting glucose is a good idea, specially for glucose-dependent tumors, the evidence shows that cancer cells also feed on glutamine. More surprinsingly, some types of cancer can grow on fatty acids (31). Restricting glucose reduces insulin levels, which promotes cancer growth. Nevertheless, ideal levels of blood glucose and insulin for treating cancer can only be achieved via calorie restriction. In fact, many supporters of ketogenic diets for cancer often cite the study of Zuccoli et al (32) on the management of glioblastoma. But very few mention what is stated in the study:
"Due to the hyperuricemia the patient was gradually shifted to a calorie restricted non-ketogenic diet, which also delivered a total of about 600 kcal/day. This diet maintained low blood glucose levels and slightly elevated (++) urine ketone levels due to the low calorie content of the diet."
Despite switching to a non-ketogenic (by definition) diet, the patient still showed progress. In my opinion, besides glucose and protein restriction, calorie restriction (and probably fasting) is the dominant factor for achieving success during cancer treatment. 

Summary and key points

  • Both energy substrates (glucose and fatty acids) support their own oxidation and inhibit the metabolism of the other.
  • A diet high in fat and low in carbohydrates will reduce glucose metabolism and increase fat metabolism. Conversely, a high carbohydrate-low fat diet increases glucose utilization and decreases fatty acid metabolism.
  • Increased glucose utilization implies upregulation of glucose membrane transporters and enzymes involved in glycolysis. Additionally, it reduces the activity of enzymes involved in fat metabolism. 
  • Increased fatty acid metabolism inhibits key glycolytic enzymes, as well as GLUT membrane translocation. It also interrupts glucose/insulin signaling and stimulates lipolytic enzymes. 
  • Chronic hyperinsulinemia, caused by peripheral insulin resistance and energy excess, aggraviates glucose intolerance. Both high glucose and high FFA levels promote this state, by different mechanisms. 
  • Under energy balance, a high fat ketogenic diet might produce muscular insulin resistance, reducing glucose tolerance. This should not be compensated by an increase in blood glucose levels. However, if energy intake exceeds calorie expenditure and/or the body utilizes predominantely FFA for energy for extendend periods of time, there can be a rise in blood glucose to non-pathological levels. This is specially relevant if there is little exercise being done (exercise promotes muscular insulin sensitivity) and/or there is an abnormal condition.
  • The etiology of glucose intolerance is very important for the proper treatment. Although calorie restriction is the primary solution for obesity/diet-induced insulin resistance, people with autoimmune (both congenital/perinatal or diet-induced)  and beta cell dysfunction should adopt a very low carb approach. 
In the end, the level of carbohydrates proposed by the Jaminet's is in the safe side. The alarmism promoted by some people is not supported. While severely restricting carbohydrates is, in my opinion, the best approach for MetSyn and obesity, once fat mass has reduced, one can tolerate more carbohydrate without problems. If the choice is restricting carbohydrates for life, you should expect a very abnormal response to any carbohydrate (being "safe" or "unsafe"). Nevertheless, lets not forget that the Perfect Health Diet is not a high carbohydrate diet, but a high-fat, low carbohydrate diet. Despite my obvious differences with Paul (33), his dietary advise is very reasonable and his diet is the first I recommend. This template, plus calorie restriction and/or fasting, is the best dietary measure one can implement. Everyone should adjust their individual carbohydrate needs, but in the end, the key is controlling and preventing inflammation. And carbohydrates per se are not inflammatory. 


*Certainly, there are people who are in the extremes of the Gaussian distribution. For these persons, extra measures should be taken. I will write about my approach in the following post.