Diabetic ketoacidosis: pathophysiology

  1. Introduction
  2. People still die from diabetic ketoacidosis. Poor patient education is probably the most important determinant of the incidence of the catastrophe that constitutes "DKA". In several series, only about a fifth of patients with DKA are first-time presenters with recently acquired Type I diabetes mellitus. The remainder are recognised diabetics who are either noncompliant with insulin therapy, or have serious underlying illess that precipitates DKA.

    Most such patients have type I ("insulin dependent", "juvenile onset") diabetes mellitus, but it has recently been increasingly recognised that patients with type II diabetes mellitus may present with ketoacidosis, and that some such patients present with "typical hyperosmolar nonketotic coma", but on closer inspection have varying degrees of ketoacidosis.

  3. Insulin and its antagonists
  4. DKA is best seen as a disorder that follows on an imbalance between insulin levels and levels of counterregulatory hormones. Put simply:

    "Diabetic ketoacidosis is due to a marked deficiency of insulin in the face of high levels of hormones that oppose the effects of insulin, particularly glucagon. Even small amounts of insulin can turn off ketoacid formation".

    Many hormones antagonise the effects of insulin. These include:

    In addition, several cytokines such as IL1, IL6 and TNF alpha antagonise the effects of insulin. [J Biol Chem 2001 Jul 13;276(28):25889-93] It is thus not surprising that many causes of stress and/or the systemic inflammatory response syndrome, appear to precipitate DKA in patients lacking insulin. Mechanisms by which these hormones and cytokines antagonise insulin are complex, including inhibition of insulin release (catecholamines), antagonistic metabolic effects (decreased glycogen production, inhibition of glycolysis), and promotion of peripheral resistance to the effects of insulin.

  5. Triggers of ketoacidosis
  6. Persons presenting with DKA are often seriously ill, not only because DKA itself is a metabolic catastrophe, but also because significant underlying infection or other disorders may be present. Common precipitants of DKA are:

  7. Deficits
  8. Patients with DKA have marked fluid and electrolyte deficits. They commonly have a fluid deficit of nearly 100ml/kg, and need several hundred millimoles of potassium ion (3-5+mmol/kg) and sodium (2-10mmol/kg), as well as being deficient in phosphage (1+ mmol/kg), and magnesium. Replacement of these deficits is made more difficult due to a variety of factors, including the pH derangement that goes with DKA. Mainly in children, an added concern is the uncommon occurrence of cerebral oedema, thought by some to be related to hypotonic fluid replacement.

    There are several mechanisms for fluid depletion in DKA. These include osmotic diuresis due to hyperglycaemia, the vomiting commonly associated with DKA, and, eventually, inability to take in fluid due to a diminished level of consciousness. Electrolyte depletion is in part related to the osmotic diuresis. Potassium loss is also due to the acidotic state, and the fact that, despite total body potassium depletion, serum potassium levels are often high, predisposing to renal losses.

  9. How diabetic keto-acidosis occurs
  10. Ketoacidosis is an extension of normal physiological mechanisms that compensate for starvation. Normally, in the fasting state, the body changes from metabolism based on carbohydrate, to fat oxidation. Free fatty acids are produced in adipocytes, and transported to the liver bound to albumin. There they are broken down into acetate, and then turned into ketoacids (acetoacetate and beta-hydroxybutyrate). The ketoacids are then exported from the liver to peripheral tissues (notably brain and muscle) where they can be oxidised. Note that during ketosis, a relatively small amount of acetone is produced, this giving ketotic patients their typical smell, often described as 'fruity'.

    DKA represents a derangement of the above mechanism. Despite vast amounts of circulating glucose, this carbohydrate cannot be used owing to lack of insulin. Ketogenic pathways are maximally "turned on", supply of ketones exceeds peripheral utilisation, and ketosis results. (There are a few other clinical states where similar keto-acidosis is seen. One is in alcoholics, who may present with marked ketosis, and a variable degree of either hypo- or mild hyperglycaemia. Another is in some pregnant women, particularly associated with hyperemesis gravidarum).

    The physiological mechanism of ketoacidosis is interesting. The rate-limiting step in the manufacture of ketones in the liver is the transfer of fatty acids (acyl groups) from Coenzyme A to carnitine. Carnitine acyl transferase I is the relevant enzyme, often referred to as CAT-I. To a certain degree, increased levels of carnitine will drive this transfer, but the main factor that inhibits CAT-I is the level of malonyl CoA in the liver. High levels of malonyl CoA effectively turn off the enzyme.

    Malonyl CoA is manufactured by another enzyme called Acetyl CoA carboxylase. Acetyl CoA carboxylase activity is in turn regulated by the amount of citric acid in the cell. The more the Krebs' cycle is whirling around (and citrate is being produced), the greater the activity of Acetyl CoA carboxylase, which in turn results in inhibition of ketoacid production. Turn off the supply of substrate into Krebs' cycle, and ketoacids are formed.

    You can work out that in the fasted state, glycolysis is diminished, the flow of substrate into the citric acid cycle drops, and ketone manufacture is turned on. This is unfortunately just what happens in diabetic ketoacidosis.

    We now understand how, in the midst of plenty, the liver cell in DKA cries 'starvation' and produces ketones! Both absence of insulin and excess glucagon result in inhibition of glycolysis. Such inhibition not only raises glucose levels, but stimulates ketone formation. Let's look in more detail at how these hormones inhibit glycolysis.

  11. Mechanisms of hyperglycaemia
  12. The marked hyperglycaemia seen associated with diabetic ketoacidosis (and that encountered in nonketotic hyperosmolar coma) is not as straightforward as was once thought! The combination of insulin lack and high glucagon levels has a variety of effects on the liver including:

    1. inhibition of glycolysis;
    2. inhibition of glycogen synthesis, (and stimulation of glycogen breakdown);
    3. stimulation of gluconeogenesis;

    Inhibition of glycolysis by glucagon

    Glucagon excess and low insulin levels both appear to have similar effects in inhibiting glycolysis. Glucagon ultimately has a potent inhibitory effect on the formation of fructose 2,6 bisphosphate . This product is very important, because it's an extremely potent allosteric regulator of a major rate-limiting enzyme in the glycolysis pathway, phosphofructokinase (often abbreviated to "PFK1"). The effect of glucagon is well characterised. When glucagon binds its cell-surface receptor, through a fairly direct G protein-receptor coupled mechanism, protein Kinase A is stimulated. Then the fun really starts, because protein kinase A phosphorylates an important regulatory enzyme called phospofructokinase 2 (PFK2 ). This latter protein is a strange duplicitous enzyme - when phosphorylated it wears one face, quite different from the unphosporylated enzyme. When phosphorylated, PFK2 acts as a phosphatase, but when un phosphorylated, it's a kinase. Phosphorylated PFK2 takes the vitally important Fructose 2,6 bisphosphate and lops off a phosphate to turn it into fructose 6 phosphate. The kinase form of PFK2 does the opposite, and results in the creation of more fructose 2,6 bisphosphate. As we hinted above, fructose 2,6 bisphosphate is a potent allosteric stimulator of the enzyme PFK1.

    The bottom line is that glucagon lowers fructose 2,6 bisphosphate levels and inhibits glycolysis; if glycolysis is inhibited, then flow of carbon atoms into the citric acid cycle slows, and ketogenesis is stimulated.

    Effects of insulin

    The insulin effect is far less well characterised, although we know that the effect is opposite to that of glucagon. It used to be thought that the main effect of insulin was mediated by a complex pathway involving a kinase called MAPK. We now know that this pathway is important in the long-term effects of insulin on cellular proliferation, but not the acute metabolic effects. The key regulator in the metabolic effects of insulin appears to be the enzyme phosphatidylinositol 3-kinase (PI3K). This in turn causes activation of a variety of kinases (atypical protein kinases C, and protein kinase B), which have profound metabolic effects, including inhibition of glycolysis and stimulation of glycogen synthesis. [J Clin Endocrinol Metab 2001 Mar;86(3):972-9; Philos Trans R Soc Lond B Biol Sci 1999 Feb 28;354(1382):485-95; Diabetes Metab 1998 Dec;24(6):477-89] Insulin raises fructose 2,6 bisphosphate levels by a mechanism that seems to depend on activation of PI3K [J Biol Chem 1996 Sep 13;271(37):22289-92].

    Note that this is not the whole story, because glucagon and insulin also have opposing effects on several other enzymes, including pyruvate kinase, and enzymes involved in glycogen synthesis/breakdown.

  13. Why patients with DKA die
  14. Death rates in DKA vary widely between published series, with death rates generally in the range of one to ten percent, although higher rates have been reported! Such variation is likely due to different reasons for presentation, and patients presenting at various stages during the evolution of DKA. Differences in management are also likely to affect outcome. Patients who are more likely to die include:

  15. Cerebral oedema in DKA
  16. As noted, DKA in children may be associated with cerebral oedema. Although uncommon (~1%), this complication may be associated with a high mortality rate (about 25% or more), and a high rate of neurological complications in survivors. The pathogenesis is far from clear. It has been noted that those who develop cerebral oedema are more likely to have a low arterial partial pressure of carbon dioxide on admission [Glaser et al]. Some studies [Krane] suggest that cerebral oedema may even be present on admission. The clinical picture in such cases is often one of initial improvement in level of consciousness, followed by gradual decline over several hours, culminating in sudden collapse, resuscitation, and an adverse outcome.

    It is often asserted that over-vigorous rehydration (especially with relatively hypotonic fluids) is the prime cause of cerebral oedema in such patients, but there is little or no evidence to support this attractive contention. Implicating relatively hypotonic fluids in the pathogenesis of this cerebral oedema is attractive because we have long known that in the face of extracellular hypertonicity, brain cells undergo complex metabolic changes. "Idiogenic osmoles" are produced in the brain to limit brain cell shrinkage. There is increased intracellular production of osmotically active substances such as myoinositol and taurine . It seems logical that rapidly administered hypotonic fluid will rush into brain cells and result in cerebral oedema. However, in experimental animals, aggressive insulin therapy is more likely to be associated with cerebral oedema than is aggressive fluid therapy! Nevertheless, current texts now generally caution one against over-vigorous fluid resuscitation in children with DKA, recommending that one replenish the fluid deficit over 36 hours or more. In addition, the old-fashioned tendency to give massive amounts of insulin is now considered unacceptable.

  17. Acidosis in DKA
  18. The initial acidosis seen with DKA is usually almost entirely attributed to elevated levels of ketoacids, which are strong anions. An equivalent way of viewing the acidosis is that it is associated with a lowered strong ion difference. Note that rapid repletion with large volumes of "unphysiological" fluid such as normal saline can be expected to worsen the degree of acidosis; conversely, there are several arguments against the use of more 'physiological' solutions such as lactated Ringer's. We explore this controversial area in a companion web page.

  19. References
  20. The following might make interesting further reading. More references are found at the end of the companion article.

    1. World J Surg 2001 Apr;25(4):461-7 Overview of glucose regulation. Tirone TA, Brunicardi FC

    2. Am J Med 2001 Dec 15;111(9):716-23 Clozapine-associated diabetes. Koller E, Schneider B, Bennett K, Dubitsky G

    3. Pharmacotherapy 2001 Nov;21(11):1448-54 Olanzapine-associated severe hyperglycemia, ketonuria, and acidosis: case report and review of literature. Seaburg HL, McLendon BM, Doraiswamy PM

    4. Glaser N, Barnett P, McCaslin I: Risk Factors for Cerebral Edema in Children with Diabetic Ketoacidosis. NEJM 2001; 344: 264-9
    5. N Engl J Med 1985 May 2;312(18):1147-51 Subclinical brain swelling in children during treatment of diabetic ketoacidosis. Krane EJ, Rockoff MA, Wallman JK, Wolfsdorf JI.
    6. Oh MS et al. Hyperchloremic Acidosis During the Recovery Phase of Diabetic Ketosis. Ann Int Med. 1978; 89: 925-7.