Control of blood glucose

This chapter is related to Section U1(iii) from the 2017 CICM Primary Syllabus, which asks the exam candidate to  "explain the control of blood glucose". Which is hilarious, because if a normal person working in the ICU were asked to explain the control of blood glucose, they would surely explain it by answering "it's controlled by the intensivist, who wants it between 5 and 10 mmol/L". Though technically accurate, that response is more in keeping with the theme of the Second Part exam revision chapter, titled "Glycaemic control among the critically ill" for obscure SEO reasons. For the First Part exam candidates, the college mainly wanted to know about the roles of endogenous insulin and glucagon, the activities of the liver as a "glucostat", and the effects of giving a bolus of 50% dextrose:

• Question 12 from the second paper of 2018 (control of blood glucose)
• Question 1 from the second paper of 2011 (50% dextrose bolus)
• Question 4 from the second paper of 2009 (control of blood glucose)
• Question 9 from the second paper of 2008 (50% dextrose bolus)
• Question 4(p.2) from the second paper of 2007 (response to hypoglycaemia)

Thus:

For something as fundamental as this one might expect there to be copious papers and review articles written about this topic, but no - it was actually much easier to find information about somatostatin receptors. "Glucose homeostasis" appears to be the search string with the greatest yield. Gerich (2000) is probably the most succinct overview of this material, and at under five pages will not overwhelm even the most flustered of exam candidates. Unfortunately, PubMed doesn't even have it as an abstract. Sohn & Ho (2020) would be another great recommendation, if only suitable for the paying subscribers of Pflügers Archiv. As is often the case, the financially unfortunate reader is left to choose between piracy and Wikipedia.

Normal range of blood glucose

"State the range of normal blood glucose" is a common expectation of the college examiners, as if demonstrating this basic med-school-level memory is somehow important for the assessment of a middle-career ICU trainee. For the record, this range has an unfairly random distribution among different regulatory bodies, as if everyone looked at a slightly different combination of population studies, but not at each others' work. Here is a handful of values for normal fasting BSL from whatever various online resources appear on the first page of Google results, as an example:

• 4.0-7.8 mmol/L (Diabetes Australia)
• 3.9-5.6 (WHO) 
• 4.0-5.4 (Diabetes UK)
• 3.9-6.0 mmol/L (MyMed.com)
• 3.0-5.4 (Royal College of Pathologists of Australasia)
• 4.0-7.0 (Royal Australian College of General Practitioners)
• 3.9‐7.8 mmol/L (American Association of Clinical Endocrinologists)
• 4.0-6.0 mmol/L (NIH StatPearls)

So. Which of these should the confused reader write in their exam paper? It seems like the Australian trainee should probably be going with whatever the Australian Diabetes Society says is normal, because it would be perilous to go against such a powerful organisation. Unfortunately, they don't seem to have settled on a normal value range, but their position statements seem to suggest that a fasting glucose > 7.0 is abnormal, that a BSL of more than 5.0 mmol/L is safe enough to drive with, and that a BSL of 3.9 falls within the lowest 5% centile for ambulatory monitoring, which gives the exam candidate something to work with. 

Is this a law of nature? Does the blood sugar level have hard physical limits that make a certain range of values mandatory to sustain multicellular life? At a risk of alienating the reader who is mainly interested in relevant information, it is probably worth noting that in the animal kingdom, at least among the vertebrates, blood glucose is correlated fairly strongly with metabolic rate. In this fashion, resting amphibian and fish BSLs are among the lowest, with the American bullfrog Rana catesbeiana operating normally at a BSL of 0.8 mmol/L. On the other range of the scale, most birds tend to enjoy being hyperglycaemic, with the domestic turkey normally having a BSL of around 11.0 mmol.L (Umminger, 1977). With increasing body size there is a subtle trend to lower BSLs , but it is not especially exciting -  a fall in average BSL from 9 mmol/L to 6 mmol/L is noted over a mass range from 10g to 10,000 kg (Kjeld and Ólafsson, 2008). As always, everything is much weirder with invertebrates. For example, crabs and molluscs tend to favour an extremely low range of BSL values (1.7-3.4 mmol/L), whereas  insects rock trehalose instead of glucose, and their blood trehalose level (BTL?) can range from 5 mmol/L to 50 mmol/L. 

Stable glucose levels during the fasting state

With that awkward digression behind us, we can now look at how the normal BSL is maintained in the normal human organism. This normal value, whatever you decided it was (let's say on average something like 4-5 mmol/L) is observed during the fasting state, otherwise known as the known as the post-absorptive state - the period of time during which you are not eating mainly because for a large proportion of it you are asleep. Because glucose is essential for the function of basically everything, there is a constant uptake of it from the blood by metabolically active tissues, and so it follows logically that there should also be a matching amount of glucose release from sites of storage and synthesis, which are mainly the liver and kidneys.  Yes, though conventional teaching of the hepatocentric patriarchy has convinced you sheeple that the liver is the main "glucostat" organ, in fact the kidneys contribute a substantial amount of glucose to the circulation, to the point where the proximal tubule can contribute about 20% of the total glucose supply inder normal postabsorptive conditions, and up to 40-50% after prolonged fasting (Legouis, 2020,  and Stumvoll et al, 1997). 

 

As this is a fasting state, and no new carbohydrate is being consumed, basically all glucose is either being synthesised de novo from amino acids and lactate, or coming from some stored glucose source (let's call it glycogen). In case you need some sort of numbers, it appears that about 50% of the glucose supply is being produced by each mechanism (Landau et al, 1996), at least under the conventional Western definition of fasting (where you're never more than fifteen minutes away from a glazed donut). When you're really fasting, gluconeogenesis gradually overtakes glycogenolysis as the main source of circulating glucose, and Landau et al were able to demonstrate that after 42 hours of starvation about 93% of the glucose supply is dependent on gluconeogenesis. Without revisiting any more material from the chapter on the physiological adaptation to starvation it will suffice to leave the reader with a confusing diagram to describe the comings and goings of major macronutrients during a day spent in a post-absorptive state:

Anyway: the point is that at any given moment, there is some obligatory glucose consumption happening, and so there needs to be a constant mobilisation of glucose to meet that demand. In this state, no storage of nutrients is occurring. The opposite is true during the postprandial state.

Stable glucose levels during the postprandial state

Considering that it takes about 5-6 hours to completely absorb the nutrients from a meal, and expecting three meals a day for most people even in this economy, we can safely say that most of the waking day is actually spent in the postprandial state. During this time there is a next influx of glucose and other nutrients. As the glucose consumption by the organ systems can be assumed to remain relatively stable, this means the total nutrient supply exceeds demand. In this scenario the excess of nutrients is stored in various sites, pto be released later during the postabsorptive state.

A single Western meal could contain up to 100g of rapidly available carbohydrate (over 550 mmol, or same as two 1L bags of 5% dextrose), and it would clearly be unsafe to dump it all into the bloodstream right away (consider what might happen if the BSL were abruptly rise to 100 mmol/L). In actual fact the postprandial blood glucose level rarely exceeds 9 mmol/L (and anything in excess of 11.1 mmol/L is considered "impaired glucose tolerance"). The main reason for this is the massive first pass effect of the liver. A large proportion of the carbohydrate load is used to replenish the hepatic glycogen reserve (perhaps 20% of the ingested glucose, to keep a an intrahepatic stash of about 400 kcal), and the excess (a further 10-20g) is sunk into fat. The rest of the glucose does end up in the systemic circulation, where it is rapidly sucked up by insulin-sensitive tissues like adipose tissue and skeletal muscle, with a rate of uptake so rapid that the BSL never increases beyond the abovementioned modest values. At the same time, the reader is reminded that this whole process takes some 3-4 hours to complete, and during that time all the glucose-hungry tissues continue consuming glucose, which helps dispose of a few more grams. In case a flowchart is somehow easier to follow , the metabolic fate of an ingested glucose load can be depicted in this reapproximation of a diagram from Gerich (2000). The original had images of the liver, skeletal muscle brain and kidneys, which have been omitted because it is assumed the readers know what those organs look like.

From the above, it follows that the systems of glucose control need to be able to:

• Sense the blood glucose level somehow
• Respond to rising glucose levels by:
  • Increasing the uptake of glucose into the liver, adipose tissue and skeletal muscle
  • Increasing the rate of glucose transformation into long term storage forms, such as fat and glycogen
  • Decreasing the rate of gluconeogenesis
• Respond to dropping glucose levels by:
  • Decreasing the rate of glucose deposition into storage forms
  • Increasing the rate of glucose mobilisation from storage forms
  • Increasing the rate of gluconeogenesis

This is the role played by insulin, glucagon, and to a lesser extent by the catecholamines.

Glucose sensing by regulatory systems

The first step is the detection of glucose concentration, which occurs at two main levels:

• Pancreatic islet cells sense glucose using glucokinase: this is a hexokinase with some unique properties which has a low affinity for glucose and which becomes active only in the middle of the normal BSL range. It is also not inhibited by its own product (glucose-6-phosphate), which means it continues working if you remain hyperglycaemia. This ensures that insulin release does not happen when you're hypoglycaemic, nor does it stop when BSL is high.
• Hypothalamic neurons sense glucose by - probably - the same sort of mechanism as pancreatic β-cells, and the most glucose-responsive populations are mainly seen in the areas of the CNS which are most involved with the autonomic nervous system  (Pénicaud et al, 2002). These may be mainly involved in the coordination of saiety and food intake, but could also directly influence the secretion of pancreatic hormones because the islets of the pancreas are well innervated with autonomic fibres. There are multiple areas of the CNS implicated in glucose homeostasis regulation, and as far as anybody can tell, the CICM does not need to know anything about them whatsoever. In case anybody is interested, they are listed in this massive table from Sohn & Ho (2020).

For the purposes of the CICM first part exam, it is only essential to recall that direct glucose sensing by the pancreatic β-cells and α-cells is the dominant mechanism.  

Physiological response to a bolus of 50% dextrose

To bring all of the abovementioned material together, as well as various sections of the chapters on insulin and glucagon, we can now create a point-form checklist of the endocrine response to changes in blood glucose. From the CICM examiner comments to Question 4(p.2) from the second paper of 2007 and Question 1 from the second paper of 2011, it seems they wanted the candidates to take a "sensors, integrators, effectors" structured approach. Trying to integrate this into an answer which one could write in under 10 minutes was challenging to say the least. Instead of writing everything that could possibly be written about this topic and presenting that as a preposterously long "model" answer, a short summary is created for the questions themselves, and a much longer and more elaborate version is presented here, with all the working shown.

Without further ado, the effect of giving 50% dextrose:

• Hyperglycaemia would transiently develop. 50ml of 50% dextrose is a hellishly concentrated sugary syrup having an osmolality of something like 2780 mOsm/kg. By administering such a highly concentrated glucose solution into the bloodstream, with lots of glucose (25g!) but little volume, one can guarantee a substantial jump in plasma glucose.
  Consider, hypothetically, if all of it was instantly distributed into the bloodstream. Those 25g of dextrose (138.9 mmol) would distribute into a total blood volume of about 5000ml (let's just inaccurately assume it's all acellular water), which currently contains 25mmol (5 mmol//L) of dextrose. Assuming it is evenly distributed, you would now have 32.4 mmol/L in every 1000ml of blood. Obviously this is not what happens in real life, as a lot of glucose would immediately be absorbed by cells under the influence of insulin (for example when Balentine et al (1998) gave a 50ml bolus of 50% dextrose to healthy non-diabetic volunteers the BSL was only about 13.6 mmol/L five minutes after the bolus). 
  This glucose would very rapidly escape from the intravascular compartment to join interstitial fluid, to say nothing of its uptake into cells, and would be mostly gone within about half an hour. To be more precise, Regittnig et al (2003) measured the transcapillary movement of radiolabeled glucose and found that 95% of it had equilibrated within about 28 minutes. 
• Serum osmolality would transiently increase. Assuming an initial osmolality of 280 (with a serum sodium of 135 and a urea of 5), we can calculate that instantly after the bolus the osmolality would increase to 307.4 mOsm/kg. The real in-vivo increase would likely be trivial, and certainly well below the osmoreceptor detection threshold (1%).
• Blood volume would transiently increase. An increase in the plasma osmolality would produce a movement of body water out of cells and into the circulating volume, analogous to the effect of mannitol. The increase, again, would be minimal, considering the rapid movement of the glucose into cells. Which brings us to:
• Glucose would diffuse into sensor cells. Specifically, into pancreatic β and α cells. They have GLUT2 transporters which are unaffected by insulin, representing a perpetually open door for glucose, which means the extracellular and the intracellular concentrations of glucose equilibrate extremely quickly. Once inside, glucokinase processes the glucose into glucose-6-phosphate, generating ATP and deactivating ATP-sensitive potassium channels, and therefore depolarising the cells. As the result:
• Glucagon secretion decreases. 
• Insulin secretion increases. The secretion is a release of granules which means a vast amount can be mobilised immediately, and the portal venous concentration of insulin rapidly becomes very high. This release is biphasic, with a high early peak, followed by a gradual slower peak:
  
• The liver is the first site of insulin action. Here, insulin begins several processes:
  • Reduced gluconeogenesis and urea synthesis
  • Reduced glycogenolysis
  • Reduced free fatty acid oxidation
  • Reduced ketone production
  • Increased glycogen synthesis
  • Increased VLDL synthesis
• Hepatic glucose uptake increases, and hepatic glucose secretion decreases, as the result of these changes. The hepatocytes do not have insulin-sensitive GLUT4 glucose transporters, and most of the increased uptake is due to the insulin-mediated increase in the activity of hexokinase which phosphorylates glucose into glucose-6-phosphate, trapping it in the cells.
• Some insulin becomes available in the peripheral circulation. Here it can influence insulin-sensitive tissues, of which the main ones are skeletal muscle and adipose tissue. 
• GLUT4 transporters are expressed on the surface of insulin-sensitive tissue cells, which increases the uptake of glucose into those cells. Additionally, insulin has several other metabolic effects on these tissues:
  • Effect of insulin on the skeletal muscle
    • Increased glycogen synthesis
    • Increased protein synthesis and decreased protein catabolism
  • Effect of insulin on adipose tissue
    • Increased synthesis of triglycerides and decreased lipolysis
    • Increased lipoprotein lipase activity
    • Increased uptake of free fatty acids
• As the glucose level decreases due to uptake, insulin secretion decreases, and glucagon secretion increases.

Physiological response to hypoglycaemia

With a decrease in blood glucose:

• Insulin secretion decreases
• Glucagon secretion increases. 
• Glucagon stimulates glucose mobilisation:
  • Hepatic glycogenolysis increases, liberating stored glucose
  • Hepatic gluconeogenesis is stimulated, mainly by the breakdown of amino acids (although the rate-limiting step here is the supply of suitable amino acids, which is not something glucagon can control). At the same time it appears to increase the activity of urea cycle enzymes, which means that ammonia levels actually decrease in spite of increased amino acid metabolism.
• Glucagon inhibits glycolysis and hepatic glycogen synthesis, which prevents the liberated glucose from being uselessly returned back into storage forms. 
• Glucagon also inhibits the synthesis of VLDLs and triglycerides, and stimulates the beta-oxidation of fatty acids in the liver, which promotes ketogenesis.

As glucose concentration increases due to the activity of glucagon, glucagon secretion is decreased, and insulin secretion is increased. The two competing influences and their mutual inhibition maintain the homeostatic setpoint and regulate the balance of glucose storage and release.