Pharmacological Treatment of Obesity

Gerald Curzon and E. Leigh Gibson

Overeating, its harmful effects and their treatment by dietary restriction and drugs are referred to in an early 16th century morality play "A Condemnation of Feasting" by Nicholas de la Chesnaye, as described by the historian John Hale (42). Dinner, Supper and Banquet invite Gluttony, Epicurism and Pleasure to eat. At Dinner they are attacked by the monsters Stone, Dropsy, Stroke and Gout. The monsters renew their assault at Supper, but Dame Experience sends her assistants, including Pill, to expel them. Supper is sentenced never to approach nearer than six hours to Dinner, and Banquet is hanged. The nature of Pill is not defined. However, in the 17th century, Culpeper (19) recommended an herb, cleavers (Galuin aparine) "to keep them lean and lank that are apt to grow fat." Another herbal material that has long been used is the seaweed bladderwrack (Fucus vesiculosus), the iodine in which presumably increases thyroxine synthesis so that the metabolism of obese hypothyroidal subjects is accelerated. Bladderwrack preparations survive (as do other herbal remedies) in the list of drugs approved for the treatment of obesity by the UK Department of Health.

Obesity may be considered a chronic pathological condition resulting from complex interactions between cultural, psychological and genetic factors. During the past 30–40 years, a markedly increased emphasis on its control has, in part, resulted from evidence of risks to the health of the obese by a spectrum of metabolic disorders, including non-insulin dependent diabetes mellitus, hypertension, hyperlipidemia, hypercholesterolemia, cardiovascular disease and gall bladder disease (49). The relationship between body weight and these conditions is not simple, largely due to the confounding effect of smoking, which is associated with low body weight (45). Nevertheless, even mild weight reduction can improve glycemic control, reduce blood pressure and cholesterol levels and increase the life span of overweight subjects (36). Weight loss may also alleviate suffering caused by social stigmatization and poor self-image. Associated pressures towards dieting have increased as a result of social changes; for example, the increasing divergence between reality and image as revealed by a comparison of Playboy centerfolds and US average weight for women of the same ages and heights (94). Snow and Harris (85) have stated that "with the demise of the girdle and corset . . . . a new way of controlling the size and shape of the female body evolved: the diet." They have illustrated this shift by numerical analysis of the changing contents of women's magazines. Despite the above influences, more people are becoming overweight or obese. Between 1976 and 1980, and 1988 and 1991, the percentages of the US population who were overweight, as defined by a body mass index (BMI = [weight in kg]/[height in m]2) of {ewc MVIMG, MVIMAGE,!greateq.bmp}27.8 for men and 27.3 for women, rose from 24.1% to 31.7% and from 26.5% to 34.9%, respectively (50). In England, obesity (as defined by BMI > 30) rose from 8% (men) and 13% (women) in 1987 to 15% and 16.5%, respectively, in 1995 (77).

A strong demand has therefore developed for treatments resulting in long-term weight loss. However, as both moralizing exhortations and non-pharmacological treatments (e.g., dietary restriction, nutritional education and psychological support) usually lead to no more than limited loss of weight (10), their supplementation by anorectic drugs is receiving much attention (see, in particular, refs. 51,69,77). Several such drugs are available, and many more are being developed, partly because the drugs that are recommended at present typically cause weight loss for only a few months. Weight then usually does not decrease further (41,68) and may rise towards (92) or even attain (37,91) baseline values during continued drug treatment. Part of the explanation for this may be a gradual loss of compliance. Resistance to anorectic drug treatment could also result from attenuation of the mechanisms controlling satiety or by which the drugs act or from reduced metabolic rate due to altered body composition (33). There are indications that this reduction may be opposed by appetite suppressants able to enhance metabolic rate (9,90). Drugs with anti-obesity properties due specifically to this effect or to effects on the absorption or metabolism of specific dietary constituents may provide new therapeutic avenues independent of appetite suppression (see below).

Drugs approved at present for the treatment of obesity in the USA and UK are listed in Table 1. However, it must be mentioned that rather conservative conclusions were reached in a recent report by the United States National Task Force on the prevention and treatment of obesity (69), i.e., that drugs "combined with appropriate behavioral approaches to change diet and physical activity, help some obese patients lose weight and maintain weight loss for at least 1 year." There is little justification for the short-term use of anorexiant medications but few studies have evaluated their safety and efficiency for more than one year. Until more data are available, pharmacotherapy cannot be recommended for routine use in obese individuals, although it may be helpful in carefully selected patients.

In this chapter, drugs that decrease feeding will usually be described as causing hypophagia (i.e., decreased feeding) in animal studies. However, unless the so-called behavioral satiety sequence is investigated, it is often unclear whether loss of appetite or another mechanism such as sedation, hyperactivity or nausea is involved (44). When the drugs are used clinically, they will be described as causing anorexia, appetite suppression or anti-obesity effects.

The serotonergic drug fenfluramine has been extensively used as an appetite suppressant. Some selective serotonin reuptake inhibitors (SSRIs) also have similar properties, and animal work has demonstrated effects on food intake of drugs acting at specific 5-hydroxytryptamine (5-HT) receptor subtypes (20,26).

The most widely prescribed appetite suppressant in the recent past has been the racemate DL-fenfluramine, either alone or together with phentermine. These treatments were largely supplanted by the D-isomer (d-fenfluramine, dexfenfluramine, DFN), which has greater potency and was claimed to have weaker side effects than the racemate. This compound has been licensed recently in most of the world (but see section on phentermine). DFN is an amphetamine derivative with an ethyl group on the side chain nitrogen and a trifluoromethyl group on the ring as shown in (Fig. 1), which also gives the structural formulae of a number of other drugs that are or have been used in the treatment of obesity. DFN differs markedly from the parent substance in its properties. In particular, it is reported to be without stimulant and abuse potential (62). The properties of DFN are described in detail in two recent accounts (24,72).

The major active metabolite of DFN is the de-ethylated derivative dexnorfenfluramine (DNFN, Fig. 1) which, in the rat, is somewhat more hypophagically potent than the parent drug (12). The pharmacokinetics of DFN are complex and exhibit large species differences (13). Kinetic parameters for obese human subjects given a single oral dose are shown in Table 2. Values did not differ significantly from those of non-obese subjects (16). The metabolite appeared more slowly in the plasma than DFN and had mean peak values about one-third those of the parent drug but remained in the circulation for much longer. The mean ratios of the areas under the curve (AUCs, DFN/DNFN) favored the parent drug, but individual values varied considerably—i.e., from 0.46 to 2.03 (10 subjects/group). In a steady-state study (64) on obese subjects given 15 mg of DFN orally twice a day (i.e., the typical clinical dosage schedule), drug and metabolite values 12 hours post dose, determined at monthly intervals over three months, were essentially constant, with means of 58 and 18 mg/L, respectively, corresponding to 0.25 and 0.09 mM (11).

Brain DFN and DNFN concentrations, especially in the hypothalamus, are probably the major determinants of its effect on food intake. Values for the sum of their concentrations are now available for 11 obese patients given 30 mg/day of DFN labeled with 18F in the trifluoromethyl group. This permits brain concentrations to be estimated in vivo by nuclear magnetic resonance (NMR). Steady-state brain levels of DFN + DNFN of 2.2. ± 0.2 mM were attained by 10 days (13). The mean steady-state plasma value determined by gas liquid chromatography was 0.25 ± 0.07 mM. The much higher brain than plasma values agree qualitatively with animal data (11).

In agreement with the release of neuronal 5-HT and inhibition of its reuptake by DFN in vitro (24) extracellular brain 5-HT is increased in vivo in rats given DFN at hypophagic dosages, as shown by hypothalamic microdialysis (75). It has usually been assumed that the hypophagic effect requires action of this extracellular 5-HT at receptors in the hypothalamus, in particular in the paraventricular nucleus (PVN) or adjacent regions. Indeed, injection of DFN, DNFN or 5-HT into the PVN all suppressed noradrenaline-induced feeding (59). However, rat experiments now strongly indicate that increased 5-HT availability is not necessary for the hypophagic response (21). For example, although DFN (2.5 mg/kg) was markedly hypophagic and significantly increased medial hypothalamic extracellular 5-HT, only the hypophagia resisted a pretreatment with p-chlorophenylalanine (PCPA), an inhibitor of 5-HT synthesis sufficient to reduce extra-raphe 5-HT to about 10% of control values (35,75). In the case of DNFN (35), PCPA pretreatment moderately increased hypophagic potency, while feeding was almost completely inhibited by a dose (1.5 mg/kg) that was without detectable effect on extracellular 5-HT, whether or not PCPA had been given (75). Furthermore, although a low dose of the selective 5-HT reuptake inhibitor fluoxetine blocked the rise of hypothalamic extracellular 5-HT caused by DFN, its hypophagic effect was unimpaired (78!popup(ch152ref78)). Hypophagia due to undetectably small amounts of 5-HT that might have escaped the action of fluoxetine was unlikely, as DFN (0.8 mg/kg) was sufficient to cause a detectable rise of extracellular hypothalamic 5-HT but was not sufficient to impair feeding.

As recently discussed (21), these results raise the question "How does DFN suppress feeding?" One possibility is that, as PCPA inhibits the synthesis of 5-HT less completely in the gastrointestinal tract than in the brain, direct action on gut 5-HT decreases appetite as a result of decreased motility. However, as the peripheral 5-HT2 antagonist xylamidine blocked the effects of dexfenfluramine on neither gut motility (80) nor food intake (70), any hypophagic effects of DFN that involve altered gut motility are probably centrally mediated.

Experiments on effects of various pretreatments have confirmed earlier evidence (5) that DFN and DNFN decrease feeding by different mechanisms. The latter but not the former substance appears to act by activating postsynaptic 5-HT2C (formerly called 5-HT1C) receptors (21,35). The DFN experiments were done shortly after giving the drug, when DNFN would have contributed little to the hypophagia (11). However, under chronic conditions (i.e., when humans are under treatment with DFN), mean steady-state plasma levels of DNFN (64) and its slightly greater hypophagic potency, compared with DFN, in the rat (12) suggest that DNFN is probably responsible, on average, for about 30–40% of the suppression of feeding after administration of the parent substance. The wide inter-subject variation of the drug/metabolite ratio found in human studies (i.e., 0.46–2.03 for areas under the curve) [16] implies that this contribution varies widely from patient to patient.

The blockade by metergoline of the direct hypophagic action of DFN suggests dependence on 5-HT receptors (5). However, DFN binds only weakly to 5-HT1A, 5-HT1B, 5-HT1D, or 5-HT2C sites (66). Further investigations on the possible involvement of other 5-HT receptors, in particular 5-HT7, are needed because the hypothalamus (a region of importance for feeding) has a particularly high concentration of 5-HT7 mRNA (63). The high-affinity, stereoselective binding of DFN and DNFN to rat, mouse and guinea pig brain is also worth noting, although these properties do not parallel the relative effects of the drugs on food intake in these species (66). Another possibility is that 5-HT1B receptors may be involved, but not via direct action of DFN. This mechanism is consistent with the effects of 5-HT antagonists on the blockade by systemic injection of DFN of feeding caused by injection of neuropeptide Y (NPY) into the PVN (39). On the other hand, in the absence of NPY injection, the selective 5-HT1B/1D receptor antagonist GR127935 was reported not to oppose DL-fenfluramine hypophagia (26).

An important outcome of further investigation of the mechanism of DFN hypophagia could be the identification of the metergoline-sensitive receptor(s) responsible for it and the development of hypophagic drugs acting selectively at such sites. The recent withdrawal of fenfluramine and DFN enhances the importance of such investigations. Other topics needing elucidation are the mechanism by which the increases of metabolic rate and thermogenesis caused by DFN occur (90) and the degree to which these increases are responsible for the effect of the drug on body weight.

In numerous clinical trials (81), DFN treatment decreased appetite in humans (3), caused obese subjects to lose weight, decreased blood pressure, improved glycemic control and had beneficial effects on plasma cholesterol, triglycerides and free fatty acids (40)). Good responders (defined as subjects losing at least 10% of initial body weight in 12 months of treatment) were predicted with high reliability by a loss of at least 2 kg in the first four weeks of treatment. Trials include a large, multi-center, double-blind study (the INDEX study) in which 822 mainly female patients with a mean overweight value of 58%, were given DFN (15 mg twice daily) or placebo in addition to a moderately calorie-restricted diet for 12 months (41). By the end of the study, 45% of placebo patients and 37% of the drug-treated group had withdrawn, mostly because of personal reasons or clinical changes, whether related to the drug or not. The greater compliance of the DFN group (P=0.002) was explicable by withdrawal of more of the placebo patients due to dissatisfaction with the smallness of their weight loss. Losses occurred only during the first six months of treatment, with significantly greater effects of drug than of placebo. When treatment was continued from six to 12 months, mean weights no longer continued to fall but rose slightly in both the drug and placebo groups, with the rise attaining significance (P<0.05) for the latter. The patients who completed the 12-month treatment had mean weight losses of 9.82 kg (DFN) and 7.15 kg (placebo), corresponding to 32% and 21% of the mean initial excess weight, respectively. However, only about 30% of the patients who were treated with the drug responded to it by a loss of weight of at least 10% which was maintained for the 12-month study period. In another trial (73) in which DFN was given to obese subjects (62% overweight, BMI = 35) for six months, the weight of those given DFN fell below that of those given placebo but was the same for both groups at five months after discontinuation of drug treatment.

Obesity is often associated with normal calorie intake in main meals but excessive snacking between meals. In agreement with a number of animal experiments, obese carbohydrate-selective snackers showed a greater decrease in snacking than obese non-selective snackers when given DFN (95). In another study (28) in which the drug was given to outpatients with moderate android obesity who overconsumed snacks of varied composition contributing 42% of their total intake, the placebo had (for presumably motivational reasons) more effect on snack intake than on main meals, and DFN caused an additional decrease in intake. Percentage reductions in total intake and of intake of dietary constituents due to the drug, compared with placebo, were similar for both main meals and snacks, the latter percentages being comparable for carbohydrates and total fats. Protein intake reductions were less marked. Thus, DFN action was not specifically manifest against carbohydrates. It is relevant that so-called carbohydrate snacks are usually also high in fat (e.g., biscuits, chocolate, etc.) and unlike previous investigations, fat intake in snacks was not kept constant. In two clinical investigations in which obese subjects were free to select foods, DFN caused selective avoidance of those with high fat content (4). An intriguing finding in the previous study (28) was that alcohol consumption between meals fell significantly in the DFN group but rose significantly in the placebo group. Further investigation of the effect of the drug on alcohol intake may therefore be worthwhile, although it should be noted that (for no obvious reason) the subjects selected for DFN drank significantly more alcohol between meals before drug treatment than those selected for placebo.

Until the recent reports of valvular lesions (18,38), DFN was thought to be well tolerated at the standard dose of 15 mg twice daily. The most frequent adverse effects were diarrhea, tiredness, dry mouth, polyuria and drowsiness (41,64), but these were usually transient. Depressive mood, as reported by the patients, was more frequent after withdrawal of DFN in the INDEX study than after withdrawal of placebo (5.1% vs. 1.5%; P<0.04) [40]. Whether this was a consequence of altered serotonergic function or of disappointment because of the weight gain that occurred is unclear. Detailed data on adverse effects, reported world-wide either spontaneously or in trials over a 10-year period, have been published (55).

Primary pulmonary hypertension, an infrequently noted but serious side effect of DFN treatment, has an onset which may be insidiously asymptomatic (79) and an incidence that increases with duration of treatment. A recent correspondence on this topic (71) recommends continued surveillance (see also section on adverse effects of phentermine).

As DFN decreases 5-HT in rat brain, there has been much discussion on whether this occurs in humans given the drug therapeutically (13,52,74). The effect was more marked in female than in male rats, a difference that was not explicable by the sex difference in pharmacokinetics that was also noted (23,74). Reductions in brain regional 5-HT of up to 24% were found in female rats seven days after a single hypophagic dose of DFN (3.8 mg/kg orally) which had (in a separate group) led to a peak brain value of DFN + DNFN of 18.6 mM. Although the relationship of male brain DFN + DNFN concentration to that of cortical 5-HT in various animal species suggests that brain 5-HT values would not be affected at the concentration of DFN + DNFN of 2 mM found in the brains of humans chronically treated with DFN (30 mg/day) [13], caution is necessary when extrapolating data from male animals to the mainly female patients who are treated with DFN. Therefore, more work on the relative effects of chronic treatment on male and female animals would be worthwhile. Nevertheless, it must be noted that major harmful effects of fenfluramine on the nervous system were apparent neither in a study of more than 3000 patients given DFN (55) nor during 25 years in which either DFN or DL-fenfluramine were given to much greater numbers of subjects.

The selective serotonin reuptake inhibitor fluoxetine (Prozac¨) is an appetite suppressant as well as an antidepressant (96). Like DFN, Prozac contains a trifluoromethyl group on the ring; unlike DFN, it is marketed as the racemate. The anorectic properties of fluoxetine have been the subject of many investigations, but, at present, it is not licensed for the treatment of obesity. It differs from most other antidepressants in that they tend to cause weight gain, though it should be noted that typical anti-obesity doses of fluoxetine are higher than those used for treatment of depression.

Fluoxetine is metabolized to the demethylated derivative norfluoxetine (Fig. 1) which has a half-life in plasma 2–3 times that of the parent drug in both rats and man. ED50 values and associated rat brain concentrations suggest that the metabolite plays an important role in the hypophagic response to the parent drug (2).

As recently reviewed (21), a number of papers strongly indicate that the hypophagic action of fluoxetine in the rat, like that of DFN, does not depend on increased extracellular 5-HT. Thus it is prevented neither by the 5-HT neurotoxin 5-7-dihydroxytryptamine (5,7-DHT) nor by PCPA. Also, rat brain concentrations of both the drug and its metabolite that are required for hypophagia are much higher than those for inhibition of 5-HT reuptake. Furthermore, evidence is against dependence of fluoxetine hypophagia on its binding to 5-HT receptors, in as much as the affinity for 5-HT1A, 5-HT1B, 5-HT1D, 5-HT2, 5-HT3 and 5-HT7 sites is weak. However, recent microstructural studies (43) indicate that metergoline opposes the hypophagic effect. NPY changes may mediate the hypophagia, as fluoxetine given either i.p. or chronically by minipump decreased NPY in the PVN and lateral hypothalamus (29). It would be of interest to know more about the contribution of the increased energy expenditure and body temperature caused by fluoxetine (9) to its effect on obesity.

In a double-blind, multi-site trial (37), fluoxetine hydrochloride (60 mg/day) or placebo was given for 52 weeks to 458 obese patients (BMI = 36) who were also advised with regard to a low-calorie diet. During the first 20 weeks, both drug and placebo subjects lost weight (-5.1 and -2.4 kg, respectively). They then gained weight—the placebo group slowly and the drug group more rapidly—so that by 52 weeks both had comparable mean weights (-1.7 kg and -2.4 k, respectively). There were major site-to-site differences perhaps related to local differences in weight reduction programs. Thus, at one site where patients were given counseling on behavior modification and encouraged to walk, the weight of the drug group continued to fall for the full 52 weeks (-13.9 kg), while that of the placebo group only fell for 20 weeks (-2.1 kg). This finding implies a need for more investigation of the interactions between drug and other treatments for obesity. Nevertheless, results as a whole do not encourage the development of fluoxetine as an anti-obesity agent.

The main adverse effects of fluoxetine (37) when given as an anti-obesity drug are headache, nausea, diarrhea and sleep disturbances but these are infrequent , usually mild and rarely sufficient to lead to discontinuation of therapy.

Another SSRI, sertraline, has been the subject of limited study as an appetite suppressant. In one investigation (67), a small group of obese patients with major depression given various doses of the drug for 8–61 weeks showed weight losses which were greater in patients with lower urinary concentrations of the norepinephrine metabolite 3-methoxy-4-hydroxyphenylglycol. However, in a 54-week, double-blind trial in which sertraline was given to obese patients after their weight had been reduced by treatment with a very low calorie diet, the drug was ineffective; the weight of the patients gradually rose so that at 26 weeks it was no different from that of a placebo group on unrestricted caloric intake (91).

As drugs with agonist activity at 5-HT2C receptors cause hypophagia in rats (35!popup(ch152ref35)), highly selective agonists at these sites have been suggested as potential anti-obesity agents (26). However, activation of these receptors also causes anxiety-like behavior (48). Therefore, drugs that activate not 5-HT2C but 5-HT1B receptors (20,26) might have more therapeutic potential, especially as the selective 5-HT1B receptor agonist CP-94,253 has been shown to enhance the natural mechanism of satiety (44). (See also the later section on bromocriptine).

Catecholaminergic agents have been the largest group of clinically used anti-obesity drugs (54) since the observation in the 1930s that the use of sympathomimetic agents such as amphetamine to treat asthma caused anorexia and weight loss (60). Various derivatives were developed (mainly phenylethylamines) in an attempt to reduce cardiovascular side-effects, and abuse potential associated with the stimulant properties. Compounds that have achieved DEA Schedule IV classification include diethylpropion, phentermine and mazindol. Although they have been widely used, especially in the USA, there is still some concern about possible abuse potential. Several other phenylethylamines that have had considerable clinical use in the past, including amphetamine, remain, somewhat surprisingly, on the list of approved drugs, albeit with Schedule II or III classification.

Clinical use of adrenergic agents in the treatment of obesity gained credence with evidence of the importance of hypothalamic noradrenergic (noradrenergic) systems for feeding behavior. However, despite the existence of a substantial preclinical literature (32), the manner in which amphetamine and related drugs affect human eating behavior remains unclear. For instance, a residential laboratory study of normal human volunteers revealed that D-amphetamine reduced the number of episodes of eating but not how much was eaten/episode and did not alter subjective hunger (31). It may be that the drug enhances preoccupation with other behaviors rather than reducing hunger as such. In recent years, concern about side-effects and abuse and competition from DL-fenfluramine and DFN have decreased the popularity of most adrenergic drugs in the treatment of obesity.

The adrenergic agent phenylpropanolamine (DL-norephedrine) has been marketed as an over-the-counter (OTC) treatment in the USA. It is readily absorbed from the gastrointestinal tract and has a half-life (t1/2b) of 3.9–4.6 hours in man (83). It does not appear to have abuse potential, but its clinical efficacy for weight loss seems also to be weak, particularly in the absence of dietary and exercise therapy, as is usually the case for an OTC drug. Adverse effects are mostly slight, but the drug is not recommended for subjects taking monoamine oxidase inhibitors (83). Initially, it was used in combination with caffeine, but this form is no longer available over the counter in the USA, mainly due to the possibility of cardiovascular side effects. However, a mixture not of phenylpropanolamine but of ephedrine with caffeine has become well established in some countries, particularly in Denmark. The rationale of adding caffeine (8) was based on evidence that it enhances the thermogenic effects of ephedrine. It also permits a lower dose of ephedrine to be used so that side effects are minimized. In a 15-week, multicenter clinical trial of obese subjects (46% overweight; BMI = 33), ephedrine/caffeine (20 mg/200 mg three times a day) was as effective in reducing weight as DFN (15 mg twice a day) [8]. Withdrawals were comparable for both drug treatments. An ephedrine/caffeine combination (plus aspirin added to oppose possible cerebrovascular complications) was found to decrease the mean weight of obese patients (initially, 65% overweight; BMI = 37) by 5.2 kg more than placebo in a small, five-month double-blind trial (22). However, blood pressure, cholesterol and glycemic control were not significantly altered.

Phentermine is an amphetamine derivative with an extra methyl group on the a-carbon atom. It has weaker sympathomimetic and stimulant properties than amphetamine. In the form of an oral, slow-release resin complex, it attains peak plasma concentrations within 8 hours and has a half-life (t1/2b) of 20–24 hours (83).

A double-blind trial of phentermine (30 mg/day), in which continuous, intermittent (1 month on, 1 month off) drug and placebo treatments were given for 36 weeks, together with dietary advice, resulted in losses of 12.2, 13.0 and 4.8 kg of body weight, respectively, which were essentially complete by 24 weeks (68). The 108 subjects who entered the trial were 54% overweight; 44 failed to complete the trial, six because of central stimulant effects. The comparable effects of continuous and intermittent treatments are intriguing, although this was not seen in a later 36-week trial in which phentermine (30 mg/day) was given continuously or intermittently (86). However, in this study, intermittent phentermine (30 mg/day) was as effective as continuous DL-fenfluramine (60 mg/day).

Combined treatment with phentermine and fenfluramine is conceptually attractive as the drugs decrease body weight by different mechanisms. A trial of phentermine + DL-fenfluramine (92) showed that very prolonged anti-obesity drug treatment is feasible. However, the striking early losses of weight were partly regained as treatment continued. Nevertheless, the so-called Fen-Phen treatment has received much unbalanced publicity in the US media and has often been too readily prescribed (51!popup(ch152ref51)). In the clinical trial, 121 patients (130–180% overweight) were given either phentermine resin (15 mg) + DL-fenfluramine hydrochloride (60 mg) or placebo each day together with periodic behavioral treatment and advice on diet and exercise. During a 28-week, double-blind period, the active drugs produced significantly more weight loss than placebo (14.2 vs. 4.6 kg, respectively; corresponding to 15.9% and 4.9% of initial weights, respectively). The losses essentially all occurred in the first 18 weeks. The drugs were then given to all subjects for 122 weeks, either continuously or intermittently, with no additional benefit. The 51 patients who remained in the study were given a final double-blind treatment for 34 weeks. By the end of the total experimental period of 184 weeks, weight losses from the initial baseline were much smaller than at 28 weeks (drug, 5 kg; placebo, 2 kg). The treatment caused decreased blood pressure and beneficial effects on cholesterol. Drug withdrawal resulted in an average regain of 2.7 kg in 20 weeks.

Phentermine + DL-fenfluramine produced ratings of drug liking in normals similar to those of D-amphetamine, which suggests possible problems due to euphorigenic effects. (6). A serious apparent side effect of the drug combination is the valvular heart disease recently described in 24 women without a previous history of heart disease (18). The Food and Drug Administration were informed of these findings before publication and issued a request for other cases to be reported. This led to the publication seven weeks later of 28 additional cases (38). Eighteen of the total of 52 patients also had pulmonary hypertension. Eleven patients required heart surgery. The FDA also received reports of valvular disease in two patients given DL-fenfluramine alone, in four patients given DFN alone, and in two patients given DFN + phentermine.

The authors of the initial report (18) comment as follows: "In the absence of a control group or a case-control study, definitive statements about a true association of valvular disease with fenfluramine-phentermine therapy cannot be made. However . . . clinically significant left-sided regurgitant valvular heart disease in a population less than 50 years old is rare. Thus, the association . . . is not likely to be due to chance. Moreover, the unusual echocardiographic morphology of the lesions further diminishes the likelihood of a coincidental observation." As the valvular lesions were similar to those seen in carcinoid disease and in ergotamine-induced valve disease, they may be mediated by 5-HT released from extracerebral stores by fenfluramine. Any adverse effect of this could be enhanced if phentermine altered the metabolism of either fenfluramine or the 5-HT released by it. A similar mechanism could explain the potentiation by phentermine of the decreases of brain regional 5-HT stores of rats given DL-fenfluramine at high dosage (61).

Diethylpropion is a phenylethylamine derivative (1-phenyl-2-diethylamine-1-propanone hydrochloride) with only slight sympathomimetic and stimulant properties. Peak plasma concentrations occur about 2 hours after oral administration. A review (83) describes three trials in which diethylpropion (75 mg/day) given for 12 weeks caused weight losses of 6–7 kg more than those obtained with placebo. In common with a phentermine trial (12), continuous and intermittent treatments were comparably effective. Adverse effects were slight (83).

The non-phenylethylamine catecholaminergic drug mazindol has moderate stimulant activity but negligible abuse potential (83). Peak plasma concentrations occur 2 hours after oral administration to humans. In animal experiments, blockade of the hypophagic effect of mazindol by a-methyl-p-tyrosine (an inhibitor of catecholamine synthesis) and the dopamine (DA) receptor blocker pimozide indicates that the action of the drug on appetite depends on dopaminergic properties (15). Short-term trials revealed that the drug caused moderately greater weight losses than placebo (83), but long-term controlled trials have not been reported (69). Some adverse effects result from the stimulant properties of mazindol, and it is contraindicated in heart disease and in conjunction with drugs likely to cause adrenergic effects (83).

Another dopaminergic drug, the D2 dopamine agonist bromocriptine, has, in a fast release formulation (Ergoset¨), given encouraging results in a small, double-blind trial (17), which somewhat unusually had a slight preponderance of male subjects (drug 5 M, 3 F; placebo 5 M, 4 F). Drug treatment for 18 weeks significantly reduced body weight and body fat vs. placebo (6.3 and 5.4 kg vs. 0.9 and 1.5 kg, respectively). In relevant experiments on genetically obese ob/ob mice (personal communication, Ergo Science), bromocriptine given for 2 weeks in conjunction with either the 5-HT precursor 5-hydroxytryptophan or the 5-HT1B agonist RU24969 had synergistic effects on both body weight and food intake. As in the clinical trial, the weight loss was largely from fat stores.

As indicated above, an important problem when catecholaminergic drugs are used to treat obesity has been how to achieve selective sympathetic arousal, so that metabolism and lipolysis are stimulated without the undesirable effects of cardiovascular stimulation. As recently reviewed (88), b-agonists have been developed which stimulate lipolysis and thermogenesis much more potently than atrial contraction (b1-receptor mediated) or inhibition of smooth muscle activity (b2-receptor mediated) and act via b3-adrenoceptors. The high concentrations of these sites in brown adipose tissue in particular has encouraged research into the anti-obesity potential of b3-adrenoceptor agonists. For example, the agonist CL 316,243, which has very high selectivity for b3 adrenoreceptors is now under investigation. Humans, unlike rodents, have little brown fat, although their white adipose tissue does contain b3 adrenoceptors—albeit at low level (88). CL314,698, the diester prodrug of the above agonist, reduced the body weights of mice and rats relative to controls largely due to a decrease of total adipose tissue (53), despite unaltered or even increased food intake. The effect appeared to involve increased thermogenesis, as when obese mice were given CL 316,243 for three weeks thermogenesis increased by 45% to reach the value for lean littermates (53).

The potential importance of b3-adrenoceptors as targets for the treatment of obesity has been highlighted by the finding of an association between a mutation in the human b3-adrenoceptor gene with early onset of non-insulin dependent diabetes mellitus and weight gain (88). However, the importance of this mutation for the development of human obesity appears to vary strikingly between racial groups.

The development of clinically efficacious b3-adrenoceptor agonists has been hindered by structural and functional differences between the human and non-human receptors. For instance, both CL 316,243 and another b3 agonist BRL 37,344 are far more effective activators of adenylate cyclase at rodent than at human b3-adrenoceptors (88). Nevertheless, this area holds some promise for the development of new pharmacological treatments of obesity.

Sibutramine hydrochloride monohydrate inhibits the reuptake of both 5-HT and NA. Its licensing for the treatment of obesity is under consideration (June 1997).

Sibutramine suppresses food intake acutely in freely feeding rats, and effects of antagonists on its action (87) suggest the involvement of both noradrenergic and serotonergic (5-HT2A/2C receptors). Synergism between enhanced availability of NA and 5-HT in the hypophagic action is supported by the synergism between the hypophagic effects of fluoxetine and nisoxetine, which inhibit 5-HT and NA uptake, respectively (87). Sibutramine is rapidly demethylated in both humans and laboratory animals. Although the primary and secondary demethylated metabolites are much more potent reuptake inhibitors than the parent compound, all three substances have similar hypophagic potencies. However, evidence that effects on 5-HT availability are not necessary for the hypophagic actions of DFN and fluoxetine (see above) indicate a need for caution with regard to the involvement of 5-HT stores in the action of sibutramine. An advantage of sibutramine treatment may be that the long half-lives of its demethylated metabolites (14–16 hours) permit steady blood levels to be achieved with a single daily dose (56).

Rats become tolerant to the hypophagic action of sibutramine after about a week of administration, but their weight continues to fall—probably because the drug stimulates thermogenesis. The effect appears to be selective for brown adipose tissue and occurs via central activation of sympathetic outflow (87).

The effects of sibutramine on the disposition of NA and 5-HT are suggestive of antidepressant activity, but weight loss rather than marked antidepressant properties were seen in early studies. In the first published trial of its effect on human body weight (93), sibutramine (5 mg/day or 20 mg/day) or placebo was given to 60 obese patients for 8 weeks. All subjects had caloric restriction, behavior therapy and increased exercise. There was a dose-dependent effect on body weight (dose and mean weight loss: placebo, 1.4 kg; 5 mg, 2.9 kg; 20 mg, 5.0 kg), but only the highest dose caused significantly more weight loss than placebo. A third of the patients on 20 mg/day reported symptoms indicative of central stimulant activity.

Other encouraging trials were subsequently reported (56), with doses of 10 or 15 mg/day producing significantly greater weight losses than placebo after as little as two weeks of treatment. In a 12-week trial with 75 patients, sibutramine (10 mg/day) was as effective as the standard dose of dexfenfluramine (15 mg twice a day) [47]. While 31% of the subjects failed to complete the study, only 7% withdrew because of adverse effects. In two longer term and larger studies of 24 weeks (N=1047; 7) and 12 months duration (N=485; 46), sibutramine (10–15 mg/day) resulted in mean weight loss of about 3–5 kg more than placebo, with the maximum loss achieved by 3–5 months. However, average drop-out rates were 35% and 47%, respectively, although in the latter study, only 13% of the subjects withdrew due to adverse effects. The drug led to improved glycemic control and beneficial effects on cholesterol (56).

The more commonly reported side-effects of sibutramine (56) include dry mouth and insomnia, probably due to inhibition of NA uptake. These effects were not considered to be important. Effects on blood pressure are of some interest, as the reduction normally associated with weight loss tended to be attenuated. Moreover, as sibutramine increased heart rate, and caution is necessary in the presence of cardiovascular disease (56).

The anti-obesity drugs now on the market (Table 2) and related drugs with similar properties derive from the accidental observation that amphetamine caused loss of weight. However, much pharmacological research on appetite is now being driven by systematic studies of the biochemistry of appetite and obesity and, as indicated below, may be leading towards new and more effective drug treatments.

Orlistat, a hydrogenated derivative of lipstatin, a lipid produced by Streptomyces toxytricini, has been recommended for approval as an anti-obesity drug in the USA and Canada (May 1997). It inhibits gastrointestinal lipases and thus reduces the absorption of fat, typically by one-third. Several clinical trials are under way. A multi-center study (27) of 188 obese subjects revealed dose-dependent weight losses, with the highest dose (120 mg three times a day) causing a mean loss of 4.74 kg in 12 weeks—significantly greater than the loss of 2.98 kg for the placebo group. The subjects were on a low-fat diet (30% by energy) providing an energy deficit of 500 kcal. Cholesterol levels were beneficially altered at the above dose. Phase 3 clinical trials are now being conducted.

One problem with the clinical use of orlistat is that excretion of unabsorbed fat results in fecal incontinence, which has been reported by 31% of patients on the above dosage. Nevertheless, drop-out rates in the trial were remarkably low (13% for placebo; 7–11% in the drug groups, with no evidence of dose-dependence), which suggests that patients are not unduly disturbed by this side effect. Indeed, evidence of excretion of undigested fat may encourage them to persist with the treatment. Higher doses, however, can cause nausea and vomiting. Reports have not yet appeared on the effects of orlistat on patients eating a more typical level of fat (e.g., 40–45% by energy). A possible problem with orlistat is its effect on the absorption of fat-soluble vitamins. In normal volunteers, orlistat at the above dosage decreased the absorption of b-carotene (97) and vitamin E but not vitamin A (65). Therefore, although marked abnormalities of vitamin A, D and E level were not seen in the clinical trial (27), vitamin supplementation may be advisable.

There has been considerable interest in the possible use in obesity of the protein leptin (OB protein), the product of the OB gene which is defective in ob/ob obese mice. Daily i.p. injection of leptin decreased food intake, body weight, fat and diabetic symptoms, and increased energy expenditure (76). Similar effects were seen for mice with diet-induced obesity, but obese mice with the presumed leptin-resistant db/db mutation were unresponsive (14). Leptin is transported to the brain by a saturable system and is thought to act there to influence the regulation of body weight. Its use in the treatment of obesity is to some degree discouraged by the fact that obese subjects have high plasma levels, but as their CSF/plasma leptin ratios are low (82), it has been suggested that reduced efficiency of delivery to the brain causes or exacerbates obesity. It is conceivable, therefore, that high doses of leptin may be beneficial and, encouraged by the results of animal experiments (14,76), clinical trials are underway.

Numerous neuropeptides have effects on food intake (57). This observation has led to recent interest in the possible use of drugs acting at peptide receptors in the control of obesity.

Attention has been paid to NPY since its intrahypothalamic injection of NPY in rats was found to elicit feeding with unsurpassed potency. Research on its role in appetite control has been encouraged by the finding that NPY occurs at high levels in ob/ob obese mice, and that when these are made deficient in NPY, their obesity and diabetes are attenuated (30). These results suggest that NPY antagonists may have anti-obesity potential. One such compound, NGD-95-1, an orally active NPY 1 receptor antagonist, is in the early stages of clinical trials by Neurogen. Another compound, SR-120819A, is undergoing preclinical studies at Sanofi. Recent evidence of the particular involvement of the NPY5 receptor in the stimulation of feeding (34) may lead to antagonists which decrease appetite but do not affect other actions of NPY.

Feeding is suppressed when the peptides cholecystokinin (1) and glucagon-like peptide -1 are injected centrally into rats. The latter compound is claimed to be the most potent known inhibitor of feeding when given by this route (89). However, animal studies in which these peptides were given by mouth do not indicate anti-obesity potential (1,89). Another peptide, galanin, increased fat intake by rats, and intrahypothalamic injection of galanin antagonists had the opposite effect (58), implying that orally effective antagonists would be candidate anti-obesity drugs.

Clinical trials of anti-obesity drugs reveal significant degrees of success but also limitations. Loss of weight is greater than that attained by non-pharmacological methods alone but usually only sufficient for a partial reversal of obesity. This outcome, though associated with significant improvements in health (36), is obviously less than ideal. As already mentioned, maximal decrease of weight occurred typically in the first six months of clinical trials and then remained almost stationary (41,46) despite continued drug treatment . Indeed, body weight often rose towards (92) or even attained (37) baseline values. Dropouts in the trials were frequent. For example, 37% and 45% of drug and placebo subjects, respectively, withdrew during a one-year trial of DFN (41). Dropouts in one-year trials of fluoxetine (37) and sibutramine (46) were 55% and 47%, respectively, with negligible differences between drug and placebo groups. In the complex trial of phentermine + dl-fenfluramine (92), the percentage of dropouts rose in essentially linear fashion with the duration of the trial (i.e., 7, 32, 55 and 57% of patients dropped out at 28, 98, 150 and 184 week, respectively). The frequency of drop-outs in clinical trials presumably varies with the nature of any concurrent psychological support and will tend to increase under non-trial conditions and when anti-obesity drugs are given for longer periods, as has recently been proposed. Poor compliance is to be expected in the chronic treatment of a condition in which patients commonly are strongly driven to a behavior, i.e., overeating, which exacerbates their symptoms. Also, motivation to persevere with treatment will tend to weaken when the initial period of weight loss is over. Conversely, perseverance is likely to be enhanced when drug efficacies are particularly striking.

These limitations imply a need for improved treatments and for the targeting of patients for treatment according to their degree of motivation to lose weight and their risk of obesity-related illness. Various criteria have been described (69) based on lower and higher BMI values according to whether obesity-related illness is present or absent. For example, labeling information for DFN recommends its use when BMI is not less than 30 kg/m2 or 27 kg/m2 in the presence of obesity—related risk factors. Other criteria include whether the obesity is of the higher risk "apple" (android) or lower risk "pear" (gynoid) types with excess weight mainly in the abdomen or hip and buttock areas, respectively. Minimum requirements for drug treatment that have also been used are percentages of body fat greater than 30 for women and 25 for men.

Anti-obesity drugs that alter aminergic mechanisms are contraindicated in patients taking other aminergic drugs such as reuptake and monoamineoxidase inhibitors. Caution is stated to be necessary (69,79) in patients with a history of major psychiatric illness, in pregnancy and lactation and in the presence of antihypertensive and hypoglycemic medication. It has also been pointed out that little is known about drug treatment of obesity in children and adolescents (25), age groups in which obesity is becoming more common. As obesity in early life often precedes adult obesity, intervention here may be particularly important.

More information on drugs in present use is needed, e.g., on effects of treatment prolonged for more than a year and on the comparative effects of concurrent and sequential administration with non-pharmacologic anti-obesity procedures. Also, more data are required on effects of drugs on body fat—amount, distribution, type—variables of importance for the development of obesity and its consequences for health (49). It should also be noted that surprisingly few data are available on the effects of chronic treatment with anti-obesity drugs on subjective or objective measures of appetite in obese patients.

It may be worthwhile to develop anti-obesity drugs with greater selectivity for specific receptors; for example, further elucidation of the sites by which DFN decreases appetite may well reveal new therapeutic avenues. Ideal anti-obesity pharmacotherapies might not only suppress appetite, especially for fat, but also oppose its deposition by decreasing its absorption or increasing its metabolism. Concurrent increase of thermogenesis would also be advantageous. As it is unlikely that a single drug would have all of these actions, treatments with combinations of drugs should be worth further investigation.

Not only the available pharmacological treatments for obesity but also non-pharmacological treatments have limited efficacy (10). While the latter are beyond the scope of this review, it should be emphasized that, at present, pharmacologic treatments are primarily intended as supplementary to procedures that provide advice on diet and exercise and psychological stimuli for long-term lifestyle changes promoting control of appetite, weight loss, and resultant reduction of health risks in clinically obese patients.

published 2000