17    The Empty Medicine Cabinet

Despite enormous investment from pharmaceutical companies and an international focus on research into obesity, today we have no drugs approved as safe and effective that have more than a small and transient effect on body weight.1 Several have been withdrawn because of concerns about safety, and many others never reached clinical use because of side effects revealed in animal tests or early human trials.

The first effective weight-loss drugs were amphetamines; by 1948, about two-thirds of weight-loss patients in the United States were being prescribed these drugs. However, they have other actions: during the Second World War, American servicemen consumed 200 million amphetamine pills to combat fatigue, and after the war amphetamines were widely used as recreational performance enhancers.

Amphetamines activate neuronal systems that use noradrenaline and dopamine as neurotransmitters. One of these is the A2 group of noradrenaline-containing neurons in the nucleus of the solitary tract in the brainstem; this site receives information from the gut via the vagus nerve, and some of the A2 neurons project to the hypothalamus to regulate food intake. Amphetamines and related drugs affect appetite by mimicking the effects of activating this pathway. However, by other actions in the hypothalamus, amphetamines affect libido, alertness, and blood pressure; their side effects include paranoia, hallucinations, and delusions, and they carry a risk of addiction.

In 1959, phentermine, a derivative of amphetamine that is much less addictive, was approved by the Food and Drug Administration (FDA) for treating obesity, but only for short-term treatment; once patients stop taking phentermine they quickly regain the lost weight.2

As well as an input from noradrenaline neurons, the hypothalamus receives an input from neurons that use serotonin as a neurotransmitter, and in the 1970s, fenfluramine, which stimulates this pathway, was approved by the FDA. In obese rodents and humans, fenfluramine reduced hunger, but produced only a small weight loss. However, whereas phentermine caused insomnia, agitation, constipation, and euphoria, fenfluramine caused drowsiness, sedation, diarrhea, and depression. At the University of Rochester, Michael Weintraub realized that, because phentermine and fenfluramine acted in different ways, they might be more potent when given together, and their opposite side effects might “cancel out.” In 1992, he published a study looking at these drugs in combination.3 Those taking the combined medication for 34 weeks lost, on average, 16% of their initial weight, while the placebo group lost just 5%. The doses that were effective—60 mg of fenfluramine and 15 mg of phentermine—were lower than the doses of these drugs when given individually.

In 1996, just four years later, 6.6 million prescriptions of the so-called fen-phen cocktail were issued in the United States. In the same year, the FDA approved dexfenfluramine, a drug closely related to fenfluramine, and this too was used mainly in combination with phentermine. The safety of these drugs in combination had not been tested: once the FDA has approved a drug, doctors in the US are free to prescribe according to their clinical judgment. The use of the two drugs in combination was “off label.”

Then, in 1996, a study in the New England Journal of Medicine warned that the incidence of pulmonary hypertension was 23 times higher in patients taking the drug combination.4 This affected just one in every 17,000 patients, but it is serious; stopping the medication might not stop the disease, and a lung or a heart-lung transplant might be needed. The following year, researchers at the Mayo Clinic reported 24 cases of a rare heart valve disease in women taking fen-phen. The FDA had already received nine similar reports, and now the agency asked all health care professionals to report such cases. This produced another 66 reports, so in 1997 the FDA asked drug manufacturers to withdraw fenfluramine and dexfenfluramine. By 1997, about 60 million people had been prescribed these drugs, mostly in Europe,5 and when fen-phen was withdrawn there was little to turn to, and a scramble ensued to find alternatives.

Fenfluramine stimulates serotonin release from neurons. When serotonin is released, its duration of action is determined by how quickly it can be recycled: transporter molecules return it to the nerve endings to be reused, and sibutramine inhibits this mechanism. Sibutramine was approved by the FDA in 1997, but by 2010 it too had been withdrawn in most countries because of safety concerns. Lorcaserin, an antagonist of the serotonin receptors expressed by appetite-regulating neurons in the arcuate nucleus, is still currently approved for treating obesity, but it has only modest effects.6

Another potential treatment emerged from studies of how cannabis acts on the brain. When cannabis became a part of hippie culture in the 1960s and ’70s, one side effect was “the munchies”—it stimulated appetite. Cannabis mimics the actions of naturally produced endocannabinoids, which are made in many neurons in the brain and which act at specific cannabinoid receptors. It became possible to make antagonists that block these receptors, and in the early twenty-first century, one such antagonist, rimonabant, was approved for use in treating obesity in Europe under the trade name Accomplia. However, in 2008 the European Medicines Agency recommended that doctors should no longer prescribe rimonabant because of the risk of serious psychiatric problems—including suicidal tendencies.7

Every current anti-obesity medication targets the brain except one. The exception is orlistat (marketed as a prescription drug under the name Xenical), which stops fats from being absorbed. In patients taking orlistat, about a third of the fat that would otherwise have been absorbed passes straight through the bowel; they need to go to the toilet more often and sometimes urgently, and they produce unpleasant fatty stools. They may also need vitamin supplements because vitamins that are soluble in fat are not properly absorbed when orlistat is taken. Unsurprisingly, there are problems with compliance: few patients tolerate the side effects for long.

To develop an effective drug with no adverse side effects is not easy. Anti-obesity drugs interfere with the normal mechanisms of appetite and metabolism, not with a disease process. The control of appetite is fundamental to our survival, and the mechanisms responsible are ancient, robust, and complex, interlinked not only with the regulation of body composition, metabolism, energy expenditure, and glucose homeostasis, but also with other fundamental drives, including the drive to reproduce. The reciprocal nature of the control of sexual behavior and appetite is conserved throughout evolution, and most drugs that affect appetite have effects on sexual behavior. It’s hard to see how this can be avoided, because the same neurons in the hypothalamus control both.

Some years ago, a pharmaceutical company invited me as an adviser to a conference on peptide effects on feeding. At a moment when my attention had faded and my energy levels were low, I was asked about studies presented that morning. As far as I could see, the only observable outcomes from the experiments reported were that the animals might eat more, or less, or the same: perhaps they might have preferred to have sex, but that choice was not on the table. It was a flippant response, but more prescient than it deserved, as many of those agents did indeed turn out to affect sexual behavior.

A good example is the case of α-MSH, the most potent inhibitor of appetite so far known.8 α-MSH is produced by the neurons of the arcuate nucleus that express the serotonin receptors that are targeted by drugs like fenfluramine, sibutramine, and lorcaserin. The α-MSH neurons spray axons to every nucleus in the hypothalamus and to most adjacent regions, and also to some distant sites in the brainstem and spinal cord. In the hypothalamus, they project to the paraventricular nucleus and supraoptic nucleus, where they contact oxytocin cells that regulate both appetite and reproductive behaviors. The axons do not go everywhere, but two types of receptor for α-MSH—MC3 and MC4 receptors—are expressed in every part of the brain.

In the 1990s, pharmaceutical companies were developing agonists for MC3 and MC4 receptors in an effort to develop new treatments for obesity. These efforts foundered in part because of an inability to influence appetite without simultaneously affecting sexual responses. When tested on male volunteers, MC4 agonists inhibited appetite but also promoted penile erection.

The names given to peptides are often disconcerting. The MSH in α-MSH stands for melanocyte-stimulating hormone. The functions of the α-MSH that is made in the brain have nothing to do with melanocytes, but α-MSH is also made in the intermediate lobe of the pituitary. As its name quite reasonably suggests, when α-MSH is secreted from the pituitary into the blood it stimulates melanocytes of the skin to produce melanin—it is the “tanning hormone.” This effect is mediated by the MC1 receptor; the gene for this receptor is polymorphic, and some of the variants are associated with red hair, fair skin, and poor tanning ability, and carry a greater risk of melanoma.9 For most of us, exposure to sunlight is a hazard that we manage with creams and common sense, but for patients with erythropoietic protoporphyria, even indoor lighting can be painful. This rare disorder arises from an enzyme deficiency that leads to the accumulation in blood of protoporphyrin, a photosensitive molecule. Melanotan I, a synthetic analog of α-MSH, is now available for these patients, for whom increased melanin offers an important protection.10

At high doses, melanotan II, another analog of α-MSH, has effects on the brain, as was discovered by Mac Hadley.11 Hadley tested, on himself, the ability of melanotan II to produce a tan, but he used what proved to be an excessive dose, a dose that affected MC3 and MC4 receptors as well as MC1 receptors. He described the consequences in an account in which we can recognize both satiety-inducing effects and effects on libido:

MTII caused a rather immediate, unexpected response: nausea and, to my great surprise, an erection (no figure provided). While I lay in bed with an emesis pan close by, I had an unrelenting erection (about 8 h duration) which could not be subdued even with a cold pack. When my wife came upon the scene, she proclaimed that I “must be crazy.” In response, I raised my arm feebly into the air and answered, “I think we may become rich.”

The behavioral effects of activating MC4 receptors—inhibition of appetite and stimulation of libido—are like those seen when oxytocin is delivered into the brain. This is no coincidence. The oxytocin cells express MC4 receptors and are targeted by axons from the α-MSH neurons, and activation of these receptors causes oxytocin release from dendrites. One site of action of this oxytocin is the ventromedial nucleus of the hypothalamus, which controls both appetite and sexual behavior: for any animal, finding food is an imperative when energy stores are low, but when this need is fulfilled the drive to reproduce becomes dominant. The α-MSH neurons are activated after eating, and they suppress appetite while stimulating sexual interest. Hunger impairs the willingness of a male rat to mate with a female, but if α-MSH is injected into his brain to mimic the effects of eating to satiation, his enthusiasm to mate is restored.11

The reciprocal regulation of appetite and sexual behavior and the roles of neuropeptides in both drives have ancient origins. The nematode Caenorhabditis elegans (C. elegans) is small and harmless; one of the simplest animals with a nervous system, and of no known economic importance. There are two sexes: a self-fertilizing hermaphrodite and a male. Of the 959 cells in a hermaphrodite, 302 are neurons, and the wiring diagram of these (the connectome) has been fully mapped; these neurons use many different signals, including about 100 different peptides. The worms move with the aid of 81 muscle cells, which generate undulations, and they find food by the smell of bacteria and from other clues to their presence, such as oxygen concentrations. C. elegans was the first multicellular organism to have its entire genome sequenced, and anyone who expected that the genome would be simple was in for a shock. The development and function of this organism is encoded by about 20,000 protein-coding genes—about the same number as in humans. At least a third of the genes have mammalian homologs, and many have remarkably conserved functions.

When crawling on a “lawn” of bacteria, C. elegans alternate between “roaming” to hunt for new food, and “dwelling” to exploit locally available food. Well-fed hermaphrodites rarely leave the local food patch, and this preference to stay depends on signaling at the npr-1 receptor, which is homologous to the mammalian NPY receptor. In mammals, NPY neurons in the arcuate nucleus are essential for appetite regulation. When energy stores are low, the NPY neurons are activated and potently inhibit the α-MSH neurons, driving feeding behavior.

In C. elegans, the switch between roaming and dwelling is regulated by serotonin neurons, which promote dwelling, and neurons that produce a neuropeptide called PDF (pigment-dispersing factor), which promote roaming.13 As already mentioned, serotonin also has important roles in appetite regulation in mammals.14 The mammalian homolog of PDF is vasoactive intestinal peptide, which controls many aspects of gut function, but is also expressed in the suprachiasmatic nucleus of the hypothalamus, where it is implicated in the circadian regulation of appetite.15

Reproductive success in C. elegans is facilitated by mate-searching behavior, which is also governed by the PDF neurons. Males will leave an abundant source of food to explore their environment if no potential mates are present—but only when fully fed. This behavior is governed by internal signals that indicate the nutritional status and the reproductive state of the male, and these are mediated by an insulin-like signaling pathway and by signaling through a steroid hormone receptor. Thus, the same set of peptides and receptors is implicated in both feeding behavior and appetite in species as different as Homo sapiens and C. elegans.

Before the human genome project, it was assumed that our hundreds of thousands of genes would yield precise targets for developing new, selective drugs that would intervene in specific ways to correct imbalances in our physiology. But we don’t have hundreds of thousands of genes; like C. elegans, we have about 20,000. Most of our genes do many different things in different cells, and, in the brain, most of our cells do many different things. Finding new drugs that are without serious side effects is hard.

In this vacuum, quack remedies from herbalists, homeopaths, naturopaths, and others abound, exploiting the desperation of patients. They don’t work, and I’m not going to waste time on them. One thing does work: gastric bypass surgery. This surgery, which reduces the volume of the stomach, is almost magically effective both in producing weight loss in the most obese patients, and unexpectedly, in correcting type 2 diabetes.16 We still don’t understand why it is so effective. Originally, it was assumed that reducing the volume of the stomach would reduce how much food was absorbed, but now it is clear that this does not explain the outcomes. The surgery alters the physiology of weight regulation by changing the signals that go from the gut to the hypothalamus. These signals control our appetite and how we regulate plasma concentrations of glucose, and they even affect our choice of foods to eat.

Gastric bypass surgery is offered only to patients for whom other interventions have failed. It has serious risks: in 2004/5, a study reported that 1 in 200 patients died from complications within six months of surgery—fen-phen was withdrawn because of complications that affected just 1 in 17,000 patients, but the regulatory regime for new drugs is much more stringent than for surgical treatments. Gastric bypass is a treatment of last resort, but its effectiveness carries an important message—that the communication between the stomach and the brain can be modified to have a beneficial impact. That message drives much of the search for new drugs.

Several new hormones that are secreted from the gut have been discovered. Among these is glucagon-like peptide 1 (GLP-1), secreted from the small intestine, which has interesting effects on glucose homeostasis as well as on food intake. The GLP-1 receptor is present in the gastrointestinal tract on the sensory endings of neurons of the vagus nerve. These neurons signal to the nucleus of the solitary tract, where GLP-1 is also expressed in neurons that project to the paraventricular nucleus: thus, there seems to be a chain of GLP-1 signaling from the gut to the hypothalamus that is responsible for effects of GLP-1 on both body weight and glucose homeostasis.17 Liraglutide, a long-acting GLP-1 analog, has been approved for the treatment of type 2 diabetes, and it produces a significant but still quite modest weight loss in adults.

Another anorectic gut hormone is peptide YY; this is secreted after a meal and it inhibits appetite by inhibiting the NPY neurons of the arcuate nucleus: these neurons have fibers that lie outside the blood-brain barrier, so peptide YY can reach them directly from the blood. Oxyntomodulin, secreted from the colon, has many effects, still incompletely understood. It reduces food intake, possibly by acting at GLP-1 receptors, but it also raises body temperature and so increases energy expenditure, and it also affects insulin secretion from the pancreas and gastric emptying.17

Because these hormones are involved specifically in communication between gut and brain, there is a reasonable hope that drugs based on them will be selective in their actions. Because peptide YY, oxyntomodulin, and GLP-1 all affect body weight but by different mechanisms, giving these in combination is likely to be more effective than any one alone, allowing lower doses of each to be used. This would reduce problems of receptor desensitization, and might also reduce the risk of adverse side effects. Steve Bloom and his colleagues at Imperial College, London, have long been studying gut hormones as potential treatments for obesity, and they have reported that the combination of peptide YY, GLP-1, and oxyntomodulin is indeed very effective for weight loss in humans.18,19

Other drugs being evaluated include agonists that act at MC4 receptors in novel ways. Peptides typically act at G protein–coupled receptors; when a peptide binds to one of these, the conformation of the receptor molecule changes, resulting in activation of a G protein that is associated with the receptor. Depending on the G protein and on other features specific to particular cells, this can result in the activation of one or more intracellular signaling pathways, with diverse effects, including on gene expression and on neuronal excitability. Because different agonists at a given receptor can produce different changes in receptor conformation, they can have different effects, and exploiting this variability is one potential way of minimizing side effects.

Another drug being used in combination therapy is naltrexone, an antagonist of opiate receptors—the receptors through which morphine and heroin act. The α-MSH neurons express these receptors, and as well as releasing α-MSH they also release β-endorphin that acts at opiate receptors as a negative-feedback signal, so blocking this receptor can enhance α-MSH release, amplifying satiety. However, opiate receptors are widely distributed in the brain, and there is natural concern about the possibility of side effects on mood.

Nobody envisages that indefinitely continued drug treatment will be a good way of treating obesity; the intention behind treatment is to restore a normal body weight but also to reset the hypothalamic appetite-regulating systems in a way that enables those systems to efficiently maintain this new body weight: how that might occur is the subject of the next chapter.

There are other things we might do about the obesity epidemic. We’ve banned smoking in restaurants and public places; we might do the same for food. Ban food advertising, ban anything that makes food tastier, close restaurants, clamp down on eating in public, sack celebrity chefs. McDonald’s made food cheap, safe, and enticing; let’s close them down. We won’t do any of those things; eating isn’t something we do in the search for happiness—it’s one of those things that is happiness. Stopping people doing things that bring them happiness generally doesn’t work, and if it does, perhaps it shouldn’t.

Notes