Ketones are by-products of fat metabolism and are produced in hepatic mitochondria 1 . More specifically, the three KB are: AcAc - if not oxidized to form usable energy, this is the source of the two other KB; acetone - not an energy source, but instead exhaled or excreted as waste; and BHB - not technically a ketone, but reversibly produced from AcAc.

The precursor for the synthesis of ketones is acetyl coenzyme A (acetyl CoA), which is formed by β-oxidation of free fatty acids in the liver. The two important determinants of ketogenesis are the availability of acetyl CoA and the mobilization of fatty acids liberated from white adipose tissue during fasting or catecholaminergic stress 2 3 .

AcAc is the central ketone body. AcAc is reduced to BHB by β-hydroxybutyrate dehydrogenase (BHD) in a NADH-requiring reversible reaction 3 :

The extent of this reaction depends on the state of the cellular pool of NAD+ 4 . When highly reduced, most or all ketones will be in the form of BHB. The ratio of BHB to AcAc reflects the redox state within the mitochondrial matrix 5 . BHB is nonvolatile and is chemically stable. This ketone body's only metabolic fate is interconversion with AcAc. Once synthesized, ketones diffuse into the circulation.

Plasma concentrations of circulating KB range from <0.1 mM/l in the postprandial state to as much as 6 mM/l during fasting, with levels reaching 25 mM/l in diabetic ketoacidosis 6 7 8 . Under normal dietary conditions, ketogenesis from fatty acids is at very low levels. This is attributable to the relatively high blood levels of the ketogenesis-inhibiting hormone insulin, and low blood levels of ketogenesis-promoting hormones, glucagon and cortisol 2 . One of the prerequisites for ketogenesis is therefore low levels of circulating insulin, as seen during starvation and diabetic ketoacidosis

The fate of circulating KB is to be taken up by extra-hepatic tissues, although 10 to 20% may be lost in the urine 8 9 . KB provide an energy source for several tissues including the brain, skeletal muscle, myocardium and liver, although the heart and kidney have the greatest capacity for utilization 10 . The rate of utilization appears to be a function of the circulating plasma levels 11 12 13 .

The major illness associated by intensive care physicians with ketosis is diabetic ketoacidosis. Outside diabetic ketoacidosis, ketone measurement is not routinely performed and has been largely used as a research tool to quantify cellular redox status. The arterial ketone body ratio (AKBR) is utilized to reflect the mitochondrial redox status of hepatic cells and may reflect cellular energy status in a number of conditions including hemorrhage, hypoxia, sepsis and hepatic resection 14 . The AKBR (ratio of AcAc to BHB) reflects the cellular NAD⁺/NADH ratio, which cannot be measured directly and is linked to the phosphorylation potential of the cell (see equation above). In critically ill patients, the entry of acetyl CoA into the tricarboxylic acid cycle is limited. Acetyl CoA consequently combines with another molecule of acetyl CoA to form acetoacetic acid, which in turn is converted to BHB leading to an altered NAD⁺/NADH ratio and a decreased AKBR 14 15 . A reduced AKBR is therefore a sign of an inefficient tricarboxylic acid cycle. The AKBR has also been used as an indirect marker of the hepatocellular functional status as the predominant site of metabolism of the ketones is in the liver.

Not surprisingly, metabolic failure at the cellular level is associated with worse clinical outcomes. A low AKBR is associated with poor outcomes in both animal and human models of illness. Yassen and colleagues measured the AKBR in 20 critically ill patients 15 . They demonstrated a progressive decrease in those patients with septic shock who died. In a study of critically ill heart failure patients, Limuro and colleagues noted a survival of only 15% of patients with AKBR <0.7 16 . Dresing and colleagues found the AKBR to be significantly lower in nonsurvivors than survivors in 30 polytrauma patients (0.6 vs. 1.3; P = 0.001) 17 . Levy and colleagues, however, failed to find a difference in AKBR levels between survivors and nonsurvivors in their cohort of septic patients 18 . In conjunction with the lactate:pyruvate ratio, the AKBR may provide further insight into the cellular redox potential of critically ill patients.

Glucose is the major fuel for the brain in humans on a balanced diet. More recently it was proposed that other substances such as lactate and pyruvate may be utilized by neurons to sustain their activity 19 . Furthermore, during times of starvation, the brain has the capacity to adapt to the use of ketones as its major energy source (and in the metabolic disease diabetes mellitus). In longstanding starvation, ketones can provide 60 to 70% of the energy needs of the brain 20 . The 1.5 kg human brain utilizes 100 to 150 g glucose per day and 20% of the total body oxygen consumption at rest 7 . Protein catabolism can supply somewhere between 17 and 32 g glucose per day, well below the minimum daily cerebral glucose requirements 21 . In fact, to supply sufficient glucose to support brain metabolic requirements from protein alone would lead to death in about 10 days instead of the 57 to 73 days 22 . Moreover in subjects undergoing total starvation for 30 days, the decrease in hunger coincides with the elevation of blood ketones to 7 mM/l. During periods of ongoing starvation, an average-size person produces about 150 g KB per day 23 . One should note, however, that ketones are incapable of maintaining or restoring normal cerebral function in the complete absence of glucose.

Astrocytes may play a significant role in the regulation of brain energy metabolism 24 . Reports suggest that astrocytes can produce KB from fatty acids and leucine and are involved with other pathways of lipid metabolism, such as lipid synthesis and lipoprotein secretion 25 26 27 28 . Furthermore, it appears that the production of lactate and KB by astrocytes may increase secondary to neurotransmitters released during enhance synaptic activity 29 30 .

Although the implications of ketone body production by astrocytes are still unclear, the fact that astrocytes outnumber neurons approximately 9:1 and occupy at least 50% of the cerebral volume suggests this constitutes a quantitatively important pathway in the brain 31 .

As previously noted, humans can achieve very high circulating KB concentrations during prolonged fasting, up to 9 mM/l (range 5.8 to 9.7 mM/l) 20 . The plasma concentration of KB is a significant factor affecting the rate of cerebral uptake. The blood-brain barrier (BBB) is relatively impermeable to most hydrophilic substances, such as KB, unless they are transported by a carrier protein. A family of proteins involved in the transport of monocarboxylic acids such as lactic acid and pyruvate across cellular membranes, called monocarboxylic acid transporters (MCT1 and MCT2), are thought to play an integral role.

It is not yet understood how MCTs are regulated. KB are transported at different rates, with the uptake of AcAc being twice that of BHB at a given arterial concentration 12 . Furthermore, prolonged fasting appears to increase the BBB uptake of ketones, with some researchers demonstrating an eightfold increase in the expression of MCT1 32 . Similarly, Hasselbalch and colleagues reported a 13-fold increase in cerebral uptake of BHB in humans following several days of fasting 33 . In contrast, rapidly increasing plasma ketone concentrations do not lead to such a significant increase in cerebral ketone uptake. Pan and colleagues noted a much smaller increase in cerebral ketone concentrations following a rapid intravenous infusion of BHB as compared with prolonged fasting, implying that MCT upregulation is partly dependent on the length of exposure to increased plasma ketone levels 34 35 .

Once across the BBB, KB enter brain cells to support cellular energy requirements. Studies of the uptake of KB by brain cells suggest entry occurs by two mechanisms: diffusion and a carrier-mediated process. MCTs have been demonstrated on neurons and glial cells although their role here appears less significant as compared with that on the BBB. The impact of mitochondrial enzyme activity involved in ketone metabolism is less clear. Certainly, mitochondrial enzymes can be upregulated by increased levels of BHB, as has been demonstrated following exposure to ketones through starvation or a prolonged high-fat diet 12 36 37 38 39 . In summary, there- fore, a prolonged high blood ketone concentration leads to an increased rate of cerebral ketone uptake and metabolism.

Data for the regional variation in ketone body metabolism are based largely on animal studies. Hawkins and colleagues and Morris noted that KB uptake varies from one brain region to another 40 41 . Following an intravenous administration of radioactive BHB to adult rats, uptake was greatest in the cerebral cortex (deep layers), in the superior and inferior colliculi, and in regions where the BBB is absent 40 41 . Areas with a low metabolic rate such as the corpus callosum and other white-matter structures demonstrated greatly reduced uptake. Interestingly, the pattern of KB uptake resembled that of glucose, except in the superficial layers of the cortex where glucose uptake was higher 42 . Similarly, MCT1 expression in the brain may resemble the pattern of KB uptake with some exceptions as MCT1 is found on glia and neurons as well as on BBB epithelial cells and because the pituitary and pineal have no BBB.

Neuroprotection is the mechanisms and strategies that, once implemented, may lead to salvage, recovery or regeneration of the nervous system. Ketones may have a neuroprotective effect (see Figure 1). Ketones represent an alternative fuel for both the normal and the injured brain 43 .

Firstly, BHB is more energy efficient than glucose and can stimulate mitochondrial biogenesis via the upregulation of genes encoding energy metabolism and mitochondrial enzymes 44 . There is a marked increase in the free energy of ATP hydrolysis, as demonstrated by a 25% increase in hydrolytic work of isolated rat perfused heart preparations when exposed to KB 45 . Also, ketones increase the intermediary metabolites delivered to the citric acid cycle with a 16-fold elevation in acetyl CoA content.

Secondly, ketones protect against glutamate-mediated apoptosis and necrosis through the attenuation of the formation of reactive oxidant species. Ketones can oxidize coenzyme Q, thus decreasing mitochondrial free radical formation. Additionally, the reduction of NAD favors reduction of glutathione, which ultimately favors destruction of hydrogen peroxide by glutathione peroxidase reaction 46 47 .

Thirdly, ketones enhance the conversion of glutamate to GABA with the subsequence enhancement of GABA- mediated inhibition 48 49 . Cerebral ketone uptake also increases cerebral blood flow 50 .

Fourthly, cell death from apoptosis is attenuated by ketone ingestion 48 . Mechanisms include reductions in activation of caspase-3, increases in calbindin and prevention of accumulation of the protein clusterin.

Lastly, cerebral ketone metabolism alters cerebral blood flow 50 . Hasselbalch and colleagues demonstrated a 39% increase in cerebral blood flow following an infusion of sodium BHB.

Adaptive changes to cerebral ketone metabolism during development and starvation are well documented. How-ever, there is evidence that other conditions including TBI, ischemia, hemorrhagic shock and hypoxia induce rapid changes in both vascular and cellular transporters that favor ketone uptake and metabolism. In addition, ketone-metabolizing enzymes demonstrate the ability to change in response to neuropathologic conditions. Collectively, these adaptations suggest that the brain may be more receptive to ketone metabolism under neuropathic conditions.

Researchers have demonstrated an increase in MCT1 and MCT2 expression following cerebral injury 51 . Similarly, ketone-metabolizing enzyme expression also appears to alter. Utilization of BHB in the brain is contingent on its conversion to AcAc by BHD, which is scarce in the adult brain, especially in the basal ganglia. Studies confirm a significant increase in BHD activity following cerebral insult 52 .

The brain's ability to utilize exogenous KB is altered. At times of oxidative stress, the ability of the brain to use glucose as a substrate for metabolic activity may become compromised. Certainly, hyperglycemia has been repeatedly shown to be deleterious to the injured brain 53 . During these times, an exogenous supply of ketones may force the brain to shift its reliance from glucose to ketones, thus taking advantage of improved transport and cellular metabolism. This shift can be advantageous in both acute brain injury models (glutamate excitotoxicity, hypoxia/ischemia, TBI) and neurodegenerative conditions (Parkinson's disease, Alzheimer's disease).

Extensive research has been performed using the glutamate-induced injury model, the cerebral hypoxia model and the ischemic injury model. Glutamate neurotoxicity is implicated in a number of neurodegenerative disorders associated with brain ischemia, hypoglycemia and cerebral trauma. During injury, glutamate and aspartate are released through the reverse action of their transport systems or via enhanced exocytotic release. This in turn leads to excitotoxicity and the production of reactive oxidant species. Administration of ketones appears to limit the extent of damage 54 55 56 57 58 . For example, Massieu and colleagues demonstrated that AcAc effectively protects against glutamate neurotoxicity both in vivo and in vitro in rats chronically treated with a glycolysis inhibitor 55 .

Hypoxic/ischemic injury leads to a decrease in oxidative metabolism by reducing oxygen availability 59 60 . Ketone administration ameliorates many of the sequelae of this process, including lactate generation, free radical damage and apoptotic cascade activation 61 62 . Suzuki and colleagues demonstrated that BHB prolonged the survival time in rat models of hypoxia, anoxia and global ischemia 63 . They further showed that BHB reduced infarct size after middle cerebral artery occlusion, irrespective of whether BHB administration was delayed.

More recently, TBI models have been developed 64 65 . TBI is associated with impaired cerebral energy production with prolonged glucose metabolic depression and reduction of ATP. Cerebral injury is further exacerbated by free radical production, and activation of poly- ADP-ribose polymerase via DNA damage 66 . As such, under these conditions of impaired glycolytic metabolism, shifting the brain toward ketone metabolism may limit the extent of cerebral injury 63 67 68 69 70 (see Table 1). Much of this work is attributable to Prins and colleagues, who demonstrated that administration of intravenous BHB after TBI in adult rats led to an increase in uptake of BHB and alleviated the decrease in ATP after injury. In a subsequent study they found that a ketogenic diet was effective in reducing the cortical contusion volume in a rat model of TBI 64 65 .

Neurodegenerative conditions may also benefit from ketone administration as there is evidence of altered cerebral metabolism. Studies of Parkinson's disease and Alzheimer's disease confirm the encouraging effects of ketones. Mechanisms are complex but include decreasing free radical production and providing acetyl CoA for energy production 45 52 71 72 73 74 . Other conditions that may be responsive to ketones include amyotrophic lateral sclerosis and certain brain tumors.

Mild ketosis is induced in humans by either fasting or the consumption of a high-fat, low-carbohydrate diet. Most studies in humans have delivered ketones via the enteral route 75 . In adults, achieving blood levels of 5 to 7 mM/l total KB requires the production or intake of about 150 g per day 23 . This level, however, is subject to the influences of glucose, insulin, cortisol and catecholamines. To achieve a therapeutic response in refractory epilepsy, it has been suggested that the level of blood ketone should be above 4 mM/l 76 .

Salts of KB can be produced. Several authors (see below) have administered sodium AcAc or sodium BHB successfully with few side effects. A new formulation called KTX 0101 consists of a combination of NaCl and BHB, although relevant concentrations are not described 77 . The main issues related to intravenous administration of ketone solutions include volume loading, sodium loading and the potential for significant metabolic alkalosis. These issues will depend on the relevant concentrations of anions and cations in the mixture and on the osmolality.

There is a paucity of research examining the effects of ketone administration in humans. Interpretation of results is further complicated by variable routes of administration, concentrations and age variability of research subjects. For instance, research on the effects of intraenous ketone preparations are confined almost exclusively to adults, whereas dietary studies include a large number of children. As far as we are aware, only two clinical studies examining the neuroprotective effects of KB in humans have been undertaken. The first by Ritter and colleagues examined the effects of a ketogenic diet on biochemical markers of metabolism in head-injury patients but did not include clinical outcomes 78 . The second study, reported by Smith and colleagues, has never been published but the authors reported on the use of a sodium BHB solution for neuroprotection during cardiac surgery 77 . The bulk of the research has taken place using the rodent model.

Some general statements, however, can be made. Studies have demonstrated that ketones can be given enterally or intravenously with minimal side effects; that utilization of KB increases with rising blood concentrations, with a plateau at approximately 128 g/24 hours; that hyperketotic states are associated with improved glucose control; that increased plasma levels lead to an increase in cerebral uptake; and that ketogenic diets may lead to improved cognitive function in patients with chronic dementia (see Table 2). Well-designed outcome studies, however, are lacking 34 35 77 78 79 80 81 82 83 84 85 .

There are minimal data on the adverse effects of ketone administration, much of which are based on chronic intake via the enteral route rather than the parenteral route. In the acute setting, effects of parenteral formulations on the pH, sodium level, lipids and glucose are likely to dominate; for example, intravenous formulations would theoretically be alkalinizing, depending on the ketone concentration. This was confirmed by Hiraide and colleagues, who noted a significant increase in pH and sodium concentrations following the intravenous administration of a 20% solution of sodium BHB to severe trauma patients 81 . Also of potential concern is the reduction in glucose cerebral metabolism and the increase in cerebral blood flow demonstrated by Hasselbalch and colleagues during administration of intravenous BHB 50 . The long-term consequences of these changes are not yet known.

Some side effects are predictable following enteral administration, such as dehydration and hypoglycemia. Other side effects are less common and present following prolonged use, including growth retardation, obesity, nutrient deficiency, decreased bone density, hepatic failure, and immune dysfunction. Freeman and colleagues reported a significant rise in mean blood cholesterol to over 250 mg/dl following prolonged intake a of ketogenic diet 86 . This effect in turn may be atherogenic, leading to lipid deposition in blood vessels. In addition there are reports of dilated cardiomyopathy in patients on the ketogenic diet, possibly as a consequence of the toxic effects of elevated plasma free fatty acids. Finally, an increased incidence in nephrolithiasis in patients on the ketogenic diet as well as increases in serum uric acid secondary have been reported 66 87 .