In the earliest phase of starvation (i.e., the postabsorptive state), hepatic glycogen is a major source of the glucose entering the circulation and remains so for 12 to 24 hours (22,23). Glucagon seems to be necessary for hepatic glycogenolysis during this period, although an increase in the level of plasma glucagon does not appear to be the primary stimulus (24,25,26). After an overnight fast, the average rate of glucose utilization by a healthy human is approximately 7 g per hour (Table 8.3) (1). By extrapolation, the 70 to 80 g of glycogen present in the liver can provide glucose to the brain and peripheral tissues for 12 to 16 hours. Two events allow the maintenance of blood glucose levels beyond this time: (a) Muscle and other tissues begin to oxidize lipid-derived fuels in place of glucose, and (b) hepatic gluconeogenesis, which is also stimulated by fatty acids, replaces glycogenolysis as the principal source of glucose entering the circulation (Fig. 8.4). As will be discussed later, glycogen breakdown in muscle does not yield significant quantities of free glucose, and after an overnight fast, gluconeogenesis by the kidney is of minor importance.

Two factors stimulate the breakdown of adipocyte triglyceride during starvation. First, the concentration of circulating insulin diminishes and, consequently, triglyceride synthesis is decreased and lipolysis is enhanced (2,27). Second, norepinephrine is released from sympathetic nerve endings and directly stimulates lipolysis by raising levels of cyclic adenosine monophosphate (cAMP) in adipocytes (2,27). Epinephrine, which is secreted from the adrenal medulla, appears to play a lesser role. The mechanisms by which FFAs are released into the circulation are discussed in the section “Adipose Tissue.” The principal users of FFAs during the early phases of starvation are skeletal muscle and liver.

Because the brain is unable to use FFAs as a fuel, it must continue to use glucose during the early phases of starvation. Gluconeogenesis is an important source of the glucose that enters the circulation even after an overnight fast (28) and becomes the major source as hepatic glycogen stores become depleted (Fig. 8.2, phase III) (23). Gluconeogenesis is responsible for approximately 35% to 60% of the hepatic glucose output after an overnight fast (12 to 15 hours postabsorptive) and for more than 97% of the output by 60 hours of starvation (23,29,30). At 60 hours, glucose production is limited not by the enzymatic capacity of the liver but by the concentration of gluconeogenic substrate in the circulation (31). During the early phases of starvation, the two principal gluconeogenic precursors are lactate and alanine (Table 8.4) (17,32,33,34).

Lactate comprises 50% of the gluconeogenic substrate of liver in a human who has fasted overnight and is the major gluconeogenic substrate throughout starvation (Table 8.4) (32). When glucose cannot be metabolized beyond pyruvate in peripheral tissues, much of the pyruvate is reduced to lactate, which is then released into the circulation (Fig. 8.5). In red blood cells and renal medulla, this reduction occurs because there are no mitochondria in which pyruvate can be oxidized. In muscle and other tissues, lactate and pyruvate are released during starvation because the activity of pyruvate dehydrogenase, the enzyme that decarboxylates pyruvate to form acetyl coenzyme A (CoA), is decreased (35). For the most part, lactate generated from glucose in this way is taken up by the liver and reconverted to glucose by the gluconeogenic pathway (33). This recycling of glucose between liver and peripheral tissues via lactate is referred to as the Cori cycle.

A second major group of gluconeogenic substrates is the amino acids. Skeletal muscle is the principal reservoir of amino acids in humans (34). During early starvation, however, the gut and liver also appear to be important sources of the amino acids entering the circulation (36). A major stimulus to protein catabolism (both decreased synthesis and increased degradation) during starvation is the decrease in plasma insulin concentrations (37,38,39). Glucagon stimulates protein degradation in liver, and glucocorticoids inhibit protein synthesis in muscle and other tissues (34). Although increases in the plasma levels of these counterinsulin hormones almost certainly play a role in modulating protein catabolism in stressful states (e.g., diabetic ketoacidosis and trauma), their concentrations are not dramatically altered during starvation, and their role here is thought to be limited.

The principal amino acids released into the circulation from muscle are alanine and glutamine (Fig. 8.6). Most of the alanine
is taken up directly by the liver, whereas glutamine is metabolized in the gastrointestinal tract, which can use it for gluconeogenesis and to generate alanine (20), and by the kidney, where it is the principal gluconeogenic substrate (19) as well as a major source of the NH₃ used for neutralizing acid in urine. Glutamine and alanine, despite comprising only 15% to 20% of muscle protein (34,39,40,41), account for 50% of the amino acids released by muscle because these amino acids can be generated from other constituents in muscle as well as from the degradation of protein. Alanine is formed by the transamination of pyruvate by alanine aminotransferase and glutamine by the amidation of glutamate by free ammonia, a reaction catalyzed by glutamine synthetase (41).

The rate of release of alanine increases markedly during starvation and other states of insulin deficiency (34,42). Despite this, the concentration of alanine in plasma is usually diminished in these situations because its uptake by the liver is stimulated to an even greater extent. Since the interorgan relationships of alanine are very much like those of lactate, a “glucose-alanine cycle” similar to the Cori cycle has been proposed (22). Impaired release of alanine from muscle has been postulated as a contributor to impaired gluconeogenesis and hypoglycemia in patients with uremia, maple syrup urine disease, and ketotic hypoglycemia of infancy and in starved women during pregnancy (43).

The other major gluconeogenic substrate is glycerol, which is derived principally from the hydrolysis of adipose tissue triglyceride. In nondiabetic subjects, the rate with which glycerol appears in the circulation correlates with adipose mass (44) and increases during starvation. Glycerol comprises about 10% of total gluconeogenic substrate during early starvation and a much higher percentage during prolonged starvation, when gluconeogenesis from amino acids is markedly diminished (see below).

With the prolongation of starvation, several events occur that limit the need for gluconeogenesis and thereby conserve body protein (Fig. 8.7). The first of these, as already noted, is an increase in the reliance of muscle and other peripheral tissues on lipid-derived fuels: initially FFAs and later both FFAs and the ketone bodies, acetoacetate and β-hydroxybutyrate. The second is a change in the fuels used by the brain. During early starvation, the CNS continues to use glucose as its exclusive fuel. However, as starvation is prolonged, plasma levels of the ketone bodies increase to values even greater than the level of glucose (Fig. 8.8). Under these circumstances, the brain, or at least parts of it, increases its use of these lipid-derived fuels (1,45,46). A third factor could be a decrease in plasma leptin, which by diminishing sympathetic nervous system activity, would diminish the basal metabolic rate.

Ketone bodies are produced from acetyl-CoA via the β-oxidation of fatty acids in the liver (Fig. 8.9). This process, termed ketogenesis, is enhanced by glucagon and inhibited by insulin. In contrast to long-chain FFAs, the ketone bodies are water-soluble and cross the blood-brain barrier via specific carrier proteins (47,48,49). Furthermore, the activity of these carriers is increased in physiologic states associated with sustained hyperketonemia such as diabetic ketoacidosis and starvation (50,51). These physiologic adaptations enhance the use of ketone bodies in place of glucose by the brain and diminish the need to degrade proteins for gluconeogenesis. It is because of
these adaptations that humans of normal weight are able to survive fasts of up to 60 to 70 days.

The decreased use of glucose by the brain during prolonged starvation is accompanied by a diminished rate of gluconeogenesis in the liver (Fig. 8.2 and Table 8.3). The latter appears to be due to decreases in protein catabolism and secondarily to the release of gluconeogenic amino acids (mostly alanine) from muscle (21,34). Some studies suggest that these adaptations in protein metabolism are related to the increased use of lipid fuels during prolonged starvation (33,52,53). Whatever the mechanism, as one proceeds from early to prolonged starvation, urinary excretion of nitrogen decreases from 12 g per day to 3 to 4 g per day, indicating a decrease in protein catabolism from 75 g per day to 12 to 20 g per day (1).

The relative contribution of the kidney to gluconeogenesis increases from 5% to 10% after an overnight fast to 50% after 3 to 4 weeks of starvation (1,53). However, in absolute amounts, renal production of glucose is still much lower than hepatic production of glucose after 1 to 2 days of fasting. The increase in renal gluconeogenesis during prolonged starvation is linked to an increase in NH₃ generation from glutamine.

Unlike amino acids, the relative importance of glycerol as a gluconeogenic precursor increases during prolonged starvation. This reflects the fact that the release of glycerol from fat is approximately 14 g per day and remains nearly constant during early and late starvation (53). After several weeks of starvation, gluconeogenesis from glycerol hypothetically provides upwards of half of the glucose oxidized by the brain.

While the adipocyte has classically been viewed as a storage depot for metabolic fuel in the form of lipid, it is now clear that the fat cell plays a central role in the endocrine regulation of energy homeostasis (57). Adipocytes not only respond to numerous hormones to regulate the storage and release of lipids but also secrete hormones (such as leptin, summarized above) that act to control energy balance and endocrine function throughout the rest of the body (57,58,59). The adipocyte also secretes a number of other protein factors, including resistin (also known as Fizz3) (60); tumor necrosis factor-α (TNF-α), an inflammatory cytokine (61); and Acrp30 (or adiponectin and adipoQ) (62), that may regulate insulin sensitivity elsewhere in the body. Although it was initially suggested that resistin increases with increasing adiposity and plays a role in insulin resistance associated with obesity, a number of subsequent studies failed to support this notion (60,63,64), and more research is necessary to determine the physiologic function of resistin. TNF-α mediates elements of insulin resistance and type 2 diabetes syndromes in some mouse models of obesity and diabetes, although the relevance of TNF-α to obesity and insulin resistance in humans remains unclear (61). In contrast to resistin and TNF-α, Acrp30 appears to mediate insulin sensitization (62). Acrp30 is an adipocyte-derived complement-related protein that is secreted as a high-order multimeric complex. Its production by adipocytes is decreased in obesity and other states of insulin resistance, and exogenously increased levels of this protein enhance numerous insulin actions. Although a great deal remains to be learned about Acrp30 (e.g., the identity of the functional proteolytic product, receptor identity) (62,65), this molecule currently commands a great deal of attention as an insulin sensitizer.