Abstract

INTRODUCTION

The cloning of the ob gene and hormone leptin has led to several new insights. The identification of leptin has uncovered a new endocrine system regulating body weight. This system provides a means by which changes in nutritional state regulate other physiologic systems. A number of leptin deficiency syndromes that are treatable with leptin replacement have been identified. The majority of obese subjects are leptin resistant, which establishes that obesity is the result of hormone resistance. Leptin treatment results in weight loss in a subset of obese patients and can also synergize with other antiobesity agents to reduce weight. Leptin provides an entry point for studying a complex human behavior. Finally, there is a powerful biological basis for obesity, a fact that is (correctly) changing public perception about this medical condition.

The ob gene was cloned in 1994, and leptin was identified in 1995 as the product of the ob gene and a hormonal signal that regulates energy balance (1–4). The passage of nearly 15 y of research and 26,880 articles on leptin provide an opportunity to assess, with some hindsight, what has been learned in the interim (S Korres, personal communication, 2008). What follows is an attempt to do so with relevant conclusions that have emerged from the cloning of the ob gene and the discovery of leptin.

CLONING OF THE ob GENE AND IDENTIFICATION OF LEPTIN

The first law of thermodynamics applies similarly to inanimate and biological systems. The elaboration of this principle by the Prussian surgeon Hermann von Helmholtz and others made it clear that, to explain the striking stability of weight in living organisms in a stable environment, organisms would need a means for maintaining a constancy of energy stores (5). For example, a remarkable stability of weight has been noted in patients monitored over long periods of times considering the large numbers of calories consumed in that interval. In addition, a precise balance between energy intake and energy expenditure has been noted over even a 2-wk time frame, which tended to average out day-to-day fluctuations in intake (6). The weight of animals that were either overfed or starved returned to that of a control group once the stimulus to alter weight was removed (2, 7–11). Finally, when animals are fed nutrients in varying dilutions, they adjust their intake to maintain constancy of the number of calories that are consumed, not the volume (12).

All these data suggested that there must be a biological system that regulates food intake and maintains homeostatic control of body weight (12). This premise was articulated often during the first half of the 20th century and was further advanced by the demonstration that body weight could be altered, in either direction, by introducing specific lesions in the hypothalamus in which ventromedial hypothalamic lesions caused obesity and lateral hypothalamic lesions caused leanness (13). These data were correctly interpreted by a number of prescient scientists as indicating that brain centers in the hypothalamus receive peripheral signals that reflect an organism's nutritional state as part of a feedback loop or loops that maintain homeostatic control of body weight (14, 15).

A key question concerned the identity of these putative signals. It was proposed that glucose, fat, and protein stores were in some way sensed, as were core temperature, recent food intake, and other variables (14, 15). The possibility that one of these signals might be derived from adipose tissue was first proposed by Kennedy although neither the nature of this factor nor the mechanism by which it acted was clear [Kennedy (16), and later Hervey (17), invoked a mechanism whereby a fat-derived factor, perhaps a steroid hormone that increased appetite, was partitioned in both the adipose tissue and the aqueous compartments and was diluted as adipose mass increased]. That this factor might be hormonal was first suggested by data from parabiosis (cross circulation) experiments between obese, ventromedial hypothalamic (VMH) lesioned and control rats (18). The observation that the normal rats paired to those with a VMH lesion ate less and lost weight suggested that the lesioned rats overproduced an appetite-suppressing factor secondary to a lesion at its site of action (ie, the hypothalamus). Although the existence of this factor could be inferred from this and other similar studies, the intrinsic difficulty of implementing a biochemical purification by using a behavioral assay of feeding behavior impeded successful efforts to identify it. In retrospect, with the identification of this factor as leptin, it is clear how challenging such a purification would be, even using modern methods, because of the requirement for chronic dosing of leptin to elicit an anorectic effect. A biochemical approach for identifying endogenous appetite suppressants is also complicated by frequent false positives that result from the observation that many compounds exert aversive effects, which nonspecifically reduce appetite.

A clue to the identity of this circulating factor was provided by data from parabiosis experiments pairing ob/ob and db/db mice to wild-type mice or to each other; ob and db are recessive mouse mutations that cause massive obesity as a result of profound hyperphagia and reduced energy expenditure. In addition, both mutants manifest a pleiotropic set of numerous other physiologic abnormalities (see below) (19). The phenotypes of both mutations are strikingly similar, which suggests that the encoded genes function in the same physiologic pathway. Results from parabiosis experiments were consistent with this possibility and further suggested that the ob gene encoded a circulating factor that suppressed food intake and body weight and that the db gene encoded its receptor. The aforementioned studies using the parabiotic union of mice with lesions of the ventromedial hypothalamus to normal rats further suggested that this receptor was localized in the hypothalamus. In aggregate, these studies were consistent with the (parsimonious) hypothesis that a circulating factor produced in adipose tissue and encoded by the ob gene acted on a receptor in the hypothalamus that was encoded by the db locus. Parabiosis experiments from genetically obese fa rats further suggested that the rat fa gene also encoded the receptor (20).

Stated more simply, the available evidence suggested that food intake and body weight—or, more precisely, adipose tissue mass—were regulated by an endocrine system. In retrospect, this conclusion seems obvious, but at the time this hypothesis was dismissed by many. The reasons for this pervasive skepticism are not entirely clear, but among them undoubtedly was the protracted time between the formulation of this hypothesis and the identification of leptin.

The introduction of a methodology that enabled the cloning of mutant genes based solely on a detailed knowledge of their position on a genetic map (positional cloning) provided a means for identifying the ob and db genes and for formally testing the aforementioned hypothesis. In 1994 the ob gene was identified by positional cloning as an ≈4.5 kb RNA that was expressed exclusively in adipose tissue (1). This RNA encoded a predicted 167–amino acid polypeptide with a signal sequence, which indicated that it was secreted and likely to circulate in plasma. The gene is disrupted in the 2 available alleles of ob; in the original C57/Bl6J ob/ob mutation, a nonsense mutation disrupts protein function, whereas in the second coisogenic ob 2j mutation, a retroviral insertion abrogrates expression of the coding sequence altogether (2, 21).

The available data at the time suggested the hypothesis that this polypeptide, now known as leptin (from the Greek root leptos, meaning “thin”), functioned as the afferent signal in a negative feedback loop that maintained stability of adipose tissue mass (1–4). If true, the following criteria had to be satisfied: leptin should circulate in plasma, its concentrations should change proportionately with increases or decreases of fat mass, and the recombinant protein should reduce food intake and body weight in lean and ob but not db mice. Finally, the db gene should encode the receptor for leptin and be localized in the hypothalamus (and possibly elsewhere).

All of these criteria were satisfied. Leptin circulates in the plasma of all mammals tested, including humans and rodents, as an ≈16-kD protein with a single disulfide bond that is required for bioactivity (2). Leptin concentrations increase with accretion of adipose tissue mass and decrease when adipose mass is lost (22). Injections or infusions of leptin reduce food intake and body weight of wild-type and ob mice but have no effect on db mice (2–4, 23). The failure of recombinant leptin to alter food intake or weight in db mice established the specificity of the effect and essentially excluded the possibility that the protein reduced weight as a result of an aversive effect (2).

The leptin receptor was identified biochemically and shown to be a cytokine family receptor that is expressed broadly (24). It was subsequently shown that leptin receptor RNA was alternatively spliced and that only one of the splice variants, referred to as ObRb at the time (also known as LepR-l), was mutant in C57Bl/Ks db/db mice (25, 26). These mutant mice show an identical phenotype to animals with null mutations of the leptin receptor or leptin itself (25, 26). This genetic evidence established the critical importance of this receptor isoform in leptin signaling. ObRb is the only receptor isoform that expresses all the protein motifs required for cytokine receptor signaling. More importantly, although the other receptor isoforms were expressed broadly, ObRb was highly enriched in the hypothalamus in precisely those nuclei that alter body weight when lesioned (25, 27). This genetic evidence suggested that leptin acted directly on the hypothalamus to regulate food intake and body weight. Consistent with a central nervous system site of action, infusions of low-dose leptin centrally replicate all the effects of peripheral leptin even at intracerebroventricular doses that do not alter plasma leptin concentrations (2). It is now known that the leptin receptor also signals at other central nervous system sites outside the hypothalamus and that its expression at these sites contributes to the broad panoply of leptin's effects (see below).

In aggregate, these data establish that leptin is a novel hormonal signal in a negative feedback loop that maintains homeostatic control of adipose tissue mass by modulating the activity of neural circuits that regulate food intake and energy expenditure. These conclusions are important not only because key elements of the homeostatic system regulating weight were identified but also because the identification of leptin and its receptor confirmed the existence of a homeostatic system, the existence of which was often debated. Although centuries of earlier work dating to Antoine Laviosier suggested that energy balance in living organisms was likely to be under homeostatic control, at the time that leptin was identified this hypothesis had become controversial in part because of the intrinsic difficulty of identifying its molecular elements. In the decades leading up to the identification of leptin, the absence of definitive evidence confirming the existence of a physiologic system regulating energy balance in the form of the molecules that compose it left a void that was filled by innumerable, largely incorrect theories about how or even if weight was regulated biologically.

CHANGES IN NUTRITIONAL STATE REGULATE OTHER PHYSIOLOGIC SYSTEMS

Leptin-deficient ob mice develop a complex phenotype that includes abnormalities in most, perhaps all, physiologic systems (28). These pervasive abnormalities are distinct from those typically manifest in human obesity, an important discrepancy that at the time the gene was cloned raised questions about whether the ob gene product was likely to be relevant for the human condition. In retrospect, it has been appreciated that all of the abnormalities of massively obese ob/ob mice (paradoxically) resemble those that develop during starvation in normal animals and humans. This apparent paradox can be most easily understood if one considers the normal response to starvation. With weight loss, fat mass is lost and leptin concentrations decrease (22). This low leptin concentration is sensed and induces a state of positive energy balance by increasing appetitive behavior and reducing energy expenditure as part of a biological response aimed at restoring fat mass. This same starvation signal (ie, low leptin) also modulates the function of other biological systems as part of the adaptive response to starvation. These responses include (but are not limited to) cessation of female ovulation, reduced immune function, a decrement in insulin signaling, alterations to homeostasis, and the development of a euthyroid sick state (29–31). A role for leptin in ovulation was consistent with the suggestion by Frisch (32) that an adipose tissue–derived factor was required to develop and maintain reproductive capacity in females. As described below, we now know that this adipose factor is leptin.

In the absence of leptin production as a result of a genetic mutation, leptin-deficient animals and humans live in a state of perceived starvation. In other words, the absence of leptin is perceived as an absence of fat. In this state, a potent signal in the form of low-plasma leptin not only induces a state of positive energy balance but also triggers additional biological responses that are normally activated only in the starved state. Despite the fact that leptin-deficient organisms become obese, the ob mutation prevents the generation of a leptin-mediated signal that would otherwise act to suppress these pathologic responses.

A key result in support of this conclusion was provided by injecting fasted animals with recombinant leptin (29). In this experiment, leptin was able to suppress the effect of starvation on ovulation, thyroid function, and other neuroendocrine responses. Analogous studies have also shown that recombinant leptin can suppress the immune abnormalities that develop in fasted animals (31).

The role of low concentrations of leptin in inducing this set of alterations in humans is confirmed by the effect of leptin treatment in many leptin-deficient states (see below). Overall, these human and animal data confirm that leptin plays a key role in the adaptive response to starvation by modulating the function of other physiologic systems. It is widely (and correctly) accepted that changes in nutritional state alter the function of other physiologic systems. It is now clear that leptin is a key means by which the change in nutritional state is communicated.

MANY LEPTIN DEFICIENCY SYNDROMES ARE TREATABLE WITH LEPTIN REPLACEMENT

Leptin mutations in humans are associated with massive obesity that is remediable by leptin treatment (33). Although leptin mutations are rare, the demonstration of a profound phenotype in these patients confirms the role of this hormone in human physiology. Note that the low incidence of leptin mutations is similar to that observed for other key hormones such as insulin; a complete loss of hormone function is catastrophic in an evolutionary context (eg, leptin-deficient humans and animals are infertile and leptin-deficient animals are likely to be more susceptible to predation) and thus strongly selected against.

Leptin-deficient patients, as mentioned above, also show a set of abnormalities in other physiologic systems and these, too, are remediated by leptin treatment (33, 34). The realization that leptin deficiency is associated with alterations in the function of other organ systems suggested the possibility that low leptin concentrations were associated with and caused other pathologic conditions. Several such conditions have been identified, the first of which was lipodystrophy.

Lipodystrophy comprises a set of conditions that are associated with the absence of or with a profound reduction in adipose tissue mass (35). Lipodystrophy can be complete or partial and is often the result of mutations in genes normally required for adipose tissue development. Lipodystrophy can also be acquired as a result of (presumed) immune alterations or, more recently, as a result of chronic HIV (36, 37). Indeed, a substantial number of HIV patients, generally those receiving triple highly active antiretroviral therapy, develop lipodystrophy.

In this condition, the reduction of adipose tissue (whatever the cause) results in the secondary deposition of lipid in other organs, particularly in the liver, and in severe, potentially intractable insulin resistance. Because of the loss of the secreting organ, leptin concentrations are pathologically low in lipodystrophic patients. The contribution of the low concentration of leptin to the consequences of lipodystrophy has been clearly established by the profound effect of leptin treatment in correcting the hepatic steatosis and insulin resistance characteristics of this condition (35, 38, 39). The response to leptin therapy was most pronounced in patients with complete lipodystrophy, but beneficial effects were also evident in some patients with partial and HIV lipodystrophy. Finally, leptin also ameliorated the neuroendocrine abnormalities that develop in patients with lipodystrophy despite the fact that a significant loss of adipose tissue mass was observed (35). This finding, and an analogous finding in patients with hypothalamic amenorrhea, uncouple the normal relation between fat mass and the response to starvation and thus confirm that it is a low leptin concentration that conveys the information that an organism is pathologically thin (not the fat mass itself).

Hypothalamic amenorrhea is another condition associated with low leptin concentrations (32). Female patients who are extremely thin show delayed puberty or sometimes fail to enter puberty at all. In addition, adult females who develop extreme leanness (often associated with extreme exercise) frequently stop menstruating (32). Women with this condition show a prepubertal pattern of gonadotrophin secretion and also manifest other neuroendocrine and metabolic abnormalities, which include premature, severe osteoporosis that in most cases cannot be treated adequately with hormone replacement therapy (40). This condition is not uncommon, affects 4–8% of women of reproductive age, and accounts for as many as 1 in 3 visits by women to an infertility clinic (C Mantzoros, personal communication, 2004). The contribution of hypoleptinemia to this condition was confirmed by the demonstration that leptin replacement therapy can restore gonadotropin pulsatility in women with hypothalamic amenorrhea, some of whom had not menstruated for years (40). Patients with hypothalamic amenorrhea also develop a set of neuroendocrine abnormalities typically associated with starvation, and those are also ameliorated by leptin treatment. In addition, there is evidence from serum markers that leptin might also improve the bone pathology that often develops in these patients, although further studies will be necessary to confirm this. Similar to the results for lipodystrophy, these patients lose adipose mass despite showing clinical improvement, thus confirming the key role of hypoleptinemia in the activation of a set of physiologic responses to starvation (41).

These data further suggest that leptin might have beneficial effects in other conditions associated with extremely low leptin concentrations. As noted by Frisch (32), delayed puberty is often associated with extreme leanness, and the failure to enter puberty in the correct temporal window can have significant, life-long consequences. Females with leptin mutations typically do not enter puberty even when showing an appropriate bone age; leptin therapy rapidly induces a pubertal pattern of gonadotrophin secretion, which reverts when leptin therapy is stopped (33). In this setting, leptin is permissive for the onset of puberty because leptin treatment does not induce puberty in young women who have not reached the appropriate bone age. Thus, it is conceivable that in some cases leptin therapy might be of benefit for inducing puberty.

Women with anorexia nervosa and with cachexia resulting from cancer or severe chronic infections also show many of the same abnormalities observed in malnourished individuals, including prominent immune abnormalities. These immune alterations often contribute to the high mortality rate of these conditions. Consistent with this, patients with leptin mutations show abnormalities in immune function, and in one extended pedigree segregating leptin mutations there was a high incidence of premature death from infectious disease (34, 42). Although leptin would have the undesirable effect of inducing weight loss in the setting of cachexia, there is evidence from animals that leptin administration can suppress these abnormalities at lower doses than are required to suppress food intake or body weight (43). Thus, it is conceivable that very-low-dose, carefully titrated leptin treatment could have beneficial effects on immune function and on other systems in cachectic patients.

MOST OBESE PATIENTS ARE LEPTIN RESISTANT

Plasma leptin concentrations in humans are highly correlated with adipose tissue mass, and most obese patients have high leptin concentrations (22). The presence of a high endogenous hormone concentration in the absence of an evident hormone effect (in this instance, leanness) suggests that there is resistance to that hormone. Thus, the initial data indicating that endogenous leptin concentrations are elevated in animal and human obesity suggested that obesity was associated with leptin resistance and that the response of obese subjects to exogenous leptin was likely to be variable (23). Leptin's efficacy was first shown to be variable in rodents in which there was a spectrum of leptin sensitivity among different strains; animals in leptin-deficient states had extreme leptin sensitivity and animals with leptin receptor mutations were the most resistant (23). Animals with diet-induced obesity produced by a cafeteria or highly palatable diet, often a reliable predictor of responses in humans, showed a minimal response to leptin administration.

A similar variability in the response to leptin was seen in obese humans. Thus, although a statistically significant effect of leptin in reducing weight was observed in a small cohort of obese patients, only a subset of obese humans (≈1/3) showed a clinically significant degree of weight loss with leptin therapy (A DePaoli, personal communication, 2001) (44). These data indicated that the utility of leptin as a monotherapy for the treatment of obesity was likely to be limited to a subset of patients. It is yet unclear whether the obese individuals who respond to leptin have lower starting leptin concentrations than nonresponders. At a given body mass index or percentage of fat, there is substantial variability of leptin and ≈10–15% of obese subjects have endogenous concentrations of leptin that are indistinguishable from those of lean patients (22). The demonstration that leptin can have therapeutic effects in some patients with low leptin concentrations in other settings (see above) suggests that leptin could potentially be efficacious in those obese patients with low plasma leptin concentrations (43). Although leptin has been shown to have potent weight-reducing effects in obese animals with low leptin concentrations, this possibility has not been directly tested in humans (43).

A key issue for future studies will be to elucidate the molecular mechanisms responsible for leptin resistance. Leptin activates signal transduction in specific neural populations in the hypothalamus and other brain regions (45). The leptin receptor signals via the Janus kinase–signal transducers and activators of transcription (JAK-Stat) pathway, which depends in part on phosphorylation of tyrosine residues of specific protein substrates, including Stat3 (45). Signaling by this class of receptors is generally time limited as a consequence of the secondary activation of SOCS (suppressor of cytokine signaling) proteins, which inhibit the JAK kinase, and of specific tyrosine phosphatases, which shut off cytokine signaling after the signal transduction pathway is initially activated. Consistent with this, the cellular effects of leptin have been amplified in cells lacking either SOCS3 or PT1b, a phosphotyrosine phosphatase. Importantly, mice with haploinsufficiency for either of these genes are resistant to obesity, remain lean on a high-fat diet, and retain leptin sensitivity (46, 47). Although these studies do not establish whether these proteins directly contribute to the development of leptin resistance in humans, these data do show that the enhancement of leptin sensitivity can protect against obesity. These results also suggest that chemical inhibitors of SOCS3 or PT1b could have potential as antiobesity agents either alone or in combination with leptin.

Recently, the hormone amylin has also been shown to ameliorate leptin resistance, although the molecular mechanism has not been fully established (48). Amylin is a peptide hormone secreted from pancreatic β cells that has a number of effects on plasma glucose that are synergistic with insulin. This agent is approved for the treatment of diabetes as an adjunct to insulin treatment. Amylin reduces glucose absorption, suppresses glucagon secretion, and reduces food intake (48).

Amylin therapy is also associated with a durable weight loss of ≈5% in humans, which establishes it as one of a number of gastrointestinal signals that regulate nutrient intake and disposition. The possibility that amylin could interact with leptin to achieve even greater weight loss was first assessed in diet-induced rats where it was shown that doses of leptin that had no discernible effect as a monotherapy could significantly enhance the response to amylin (48). The data further suggested that there was true synergy between leptin and amylin and that amylin could restore leptin signal transduction in the hypothalamus of DIO (diet-induced obesity) rats. The clinical significance of these findings was also assessed in humans in which the combination of leptin and amylin resulted in a substantial weight loss of 12.9%, which was significantly greater than that of either agent alone (48). Further studies will reveal whether this combination is safe and efficacious for the treatment of obesity.

LEPTIN PROVIDES AN ENTRY POINT FOR STUDYING A COMPLEX HUMAN BEHAVIOR

Leptin treatment of leptin-deficient animals or humans has extremely potent effects on food intake. In one instance, the food intake of a leptin-deficient child was monitored before and after the first series of leptin injections. Before the first injection, this then 3-y-old child consumed at least 2000 kcal at a single test meal (S O'Rahilly, personal communication, 2001). This is the approximate number of calories an adult might eat in an entire day. After a short period of leptin treatment, this child consumed 180 kcal at an identical test meal. Marked effects of leptin in reducing food intake in lipodystrophic patients and others have also been noted (35).

The mechanism by which leptin reduces food intake has been partially elucidated. Leptin acts by modulating the activity of a set of neural pathways and, in general, activates pathways that inhibit food intake and inhibits pathways that activate food intake. A portion of leptin's actions appear to result from inhibition of neuropeptide Y/Agouti gene-related peptide (NPY/AGRP)–expressing neurons in the arcuate nucleus of the hypothalamus. The NPY/AGRP neurons act to stimulate appetite. Leptin also activates pro-opiomelanocortin (POMC) neurons that reduce appetite by activating the MC4 receptor in other parts of the brain (49). It is likely that other neural pathways also play a role, and it is not clear which aggregate set of pathways can account for the full effect of leptin treatment. Moreover, it is also not clear whether leptin resistance abrogates leptin signaling by the NPY or POMC neurons or by as-yet-unknown neural circuits. The identification of those circuits that are altered in the leptin-resistant state would provide key entry points for delineating the molecular mechanisms.

It is also unknown which of the sites to which these and other hypothalamic neurons project are responsible for the reduction of food intake. It has recently become clear, however, that leptin signaling has important modulatory effects on dopaminergic neurons that are part of reward pathways that provide input to the nucleus accumbens (50, 51). Consistent with this, functional magnetic resonance imaging of leptin-deficient subjects has revealed a marked increase in activity in the nucleus accumbens in response to images of food even after eating (52–54). Normal individuals typically show this activity only in the starved state and do not show this response after eating. The activity that is seen in the accumbens in leptin-deficient patients after eating was normalized by the use of leptin therapy (52–54). In addition, leptin-deficient patients show markedly different food preferences and perception of food than do normal individuals (52–54). These data establish that leptin plays an important role in regulating feeding, a typical complex motivational behavior. Further studies of the neural circuit activated by leptin may provide a means for elucidating the ways by which complex behavioral decisions are generated.

BIOLOGICAL BASIS FOR OBESITY

The importance of the aforementioned neural circuits for human obesity has been established in a series of genetic studies in which it has been shown that single-gene Mendelian defects account for ≈8% (or perhaps more) of cases of morbid human obesity (55–63). In one cohort, 15% of individuals with severe, early-onset obesity were members of consanguineous pedigrees (S O'Rahilly, personal communication, 2005). Thus, in addition to mutations in the leptin and leptin receptor genes, significant obesity is also demonstrable in patients with POMC and MC4R mutations. This level of Mendelian inheritance exceeds that for most, perhaps all, other complex traits, all of which are commonly accepted to have a genetic basis. Consistent with this, twin studies and other approaches have established a substantial genetic basis for obesity with a heritability of 0.7–0.9, a level exceeded only by height (64). Although a subset of these individuals shows Mendelian inheritance, the majority of these cases are likely to be the result of polygenes interacting with environmental factors.

The identification of genetic factors that cause obesity is beginning to have an impact on the public perception of the obese. In the United States, the obese are too often held to be responsible for their condition. Obese individuals on average earn less money than their no-better-qualified lean counterparts and are promoted less readily (65, 66). In addition, the obese are often the object of ridicule to an extent that would be unacceptable for any other trait.

The identification of human mutations that cause obesity requires that we modify the explanation that is often invoked to explain the pathogenesis of obesity: “The obese eat too much and exercise too little.”Although this is undoubtedly true, the deeper question is, “Why do the obese eat more and exercise less?”The answer appears to be less about the conscious choices that the obese make and more about their biological makeup. The identification of leptin and other components in a physiologic system that maintains energy balance has established that feeding is at its core a basic biological drive analogous to thirst, breathing, and reproduction (65). Although one can consciously override a basic drive over the short term, over time the basic drive to eat dominates. There are innumerable instances in our common experience suggesting that basic drives typically trump a conscious desire. The world would be a better place if people who deride the obese kept this in mind. (Other articles in this supplement to the Journal include references 67–70).

Acknowledgments

REFERENCES