Few medicines, in the history of pharmaceuticals, have been greeted with as much exultation as a green-and-white pill containing 20 milligrams of fluoxetine hydrochloride — the chemical we know as Prozac. In her 1994 book "Prozac Nation," Elizabeth Wurtzel wrote of a nearly transcendental experience on the drug. Before she began treatment with antidepressants, she was living in "a computer program of total negativity . . . an absence of affect, absence of feeling, absence of response, absence of interest." She floated from one "suicidal reverie" to the next. Yet, just a few weeks after starting Prozac, her life was transformed. "One morning I woke up and really did want to live. . . . It was as if the miasma of depression had lifted off me, in the same way that the fog in San Francisco rises as the day wears on. Was it the Prozac? No doubt."
Like Wurtzel, millions of Americans embraced antidepressants. In 1988, a year after the Food and Drug Administration approved Prozac, 2,469,000 prescriptions for it were dispensed in America. By 2002, that number had risen to 33,320,000. By 2008, antidepressants were the third-most-common prescription drug taken in America.
Fast forward to 2012 and the same antidepressants that inspired such enthusiasm have become the new villains of modern psychopharmacology — overhyped, overprescribed chemicals, symptomatic of a pill-happy culture searching for quick fixes for complex mental problems. In "The Emperor's New Drugs," the psychologist Irving Kirsch asserted that antidepressants work no better than sugar pills and that the clinical effectiveness of the drugs is, largely, a myth. If the lodestone book of the 1990s was Peter Kramer's near-ecstatic testimonial, "Listening to Prozac," then the book of the 2000s is David Healy's "Let Them Eat Prozac: The Unhealthy Relationship Between the Pharmaceutical Industry and Depression."
In fact, the very theory for how these drugs work has been called into question. Nerve cells — neurons — talk to one another through chemical signals called neurotransmitters, which come in a variety of forms, like serotonin, dopamine and norepinephrine. For decades, a central theory in psychiatry has been that antidepressants worked by raising serotonin levels in the brain. In depressed brains, the serotonin signal had somehow been "weakened" because of a chemical imbalance in neurotransmitters. Prozac and Paxil were thought to increase serotonin levels, thereby strengthening the signals between nerve cells — as if a megaphone had been inserted in the middle.
But this theory has been widely criticized. In The New York Review of Books, Marcia Angell, a former editor of The New England Journal of Medicine, wrote: "After decades of trying to prove [the chemical-imbalance theory], researchers have still come up empty-handed." Jonathan Rottenberg, writing in Psychology Today, skewered the idea thus: "As a scientific venture, the theory that low serotonin causes depression appears to be on the verge of collapse. This is as it should be; the nature of science is ultimately to be self-correcting. Ideas must yield before evidence."
Is the "serotonin hypothesis" of depression really dead? Have we spent nearly 40 years heading down one path only to find ourselves no closer to answering the question how and why we become depressed? Must we now start from scratch and find a new theory for depression?
Science may be self-correcting, but occasionally it overcorrects — discarding theories that instead need to be rejuvenated. The latest research suggests that serotonin is, in fact, central to the functioning of mood, although its mechanism of action is vastly more subtle and more magnificent than we ever imagined. Prozac, Paxil and Zoloft may never turn out to be the "wonder drugs" that were once advertised. But they have drastically improved our understanding of what depression is and how to treat it.
Our modern conception of the link between depression and chemicals in the brain was sparked quite by accident in the middle of the last century. In the autumn of 1951, doctors treating tubercular patients at Sea View Hospital on Staten Island with a new drug — isoniazid — observed sudden transformations in their patients' moods and behaviors. The wards — typically glum and silent, with moribund, lethargic patients — were "bright last week with the happy faces of men and women," a journalist wrote. Patients laughed and joked in the dining hall, as if a dark veil of grief had lifted. Energy flooded back and appetites returned. Many, ill for months, demanded five eggs for breakfast and then consumed them with gusto. When Life magazine sent a photographer to the hospital to investigate, the patients could no longer be found lying numbly in their beds: they were playing cards or dancing in the corridors.
If the men and women at Sea View were experiencing an awakening, then a few hundred miles south, others at Duke's hospital encountered its reverse. In 1954, a 28-year-old woman was prescribed Raudixin to control her blood pressure. A few months later, she returned to the hospital, complaining of crying spells, dullness and lethargy. She felt futile, guilty and hopeless, she told her doctors. A few months later, when she returned, the sense of futility had turned into hostility. A 42-year-old woman prescribed Raudixin told her doctor that "God would cause her to become insane" before she could repent. The "feeling blue," as another patient described it, persisted until the drug was discontinued. At another hospital, one patient treated with Raudixin attempted suicide. Several people had to be admitted to psychiatric wards and administered electroconvulsive therapy before the symptoms were alleviated.
Psychiatrists and pharmacologists were quick to note these bizarre case reports. How, they wondered, could simple, seemingly unrelated chemicals like Raudixin or isoniazid produce such profound and opposite effects on mood? It was around this same time that scientists were learning that the brain itself was immersed in a soup of chemicals. In the early part of the century, scientists wondered how nerve cells talked to one another. By the late 1960s, evidence suggested that signals between neurons were carried by several chemicals, including the neurotransmitter serotonin. Might isoniazid and Raudixin have altered the levels of some neurotransmitters in the brain, thereby changing brain signaling and affecting mood? Strikingly so, scientists found. Raudixin — the "feeling blue" drug — drastically lowered the concentration of serotonin and closely related neurotransmitters in the brain. Conversely, drugs known to increase euphoria, like isoniazid, increased those levels.
These early findings led psychiatrists to propose a radical new hypothesis about the cause and treatment of depression. Depression, they argued, was a result of a "chemical imbalance" of neurotransmitters in the brain. In the normal brain, serotonin shuttled between mood-maintaining neurons, signaling their appropriate function. In the depressed brain, this signal had somehow gone wrong. The writer Andrew Solomon once evocatively described depression as a "flaw in love" — and certainly, the doctors using Raudixin at Duke had seen that flaw emerge grimly in real time: flaws in self-love (guilt, shame, suicidal thoughts), love for others (blame, aggression, accusation), even the extinction of a desire for love (lethargy, withdrawal, dullness). But these were merely the outer symptoms of a deeper failure of neurotransmitters. The "flaw in love" was a flaw in chemicals.
Powerful vindication for this theory came from the discovery of new medicines that specifically elevated serotonin concentrations. The first such drug, Zimelidine, was created by a Swedish researcher, Arvid Carlsson. Following Carlsson's lead, pharmaceutical chemists threw their efforts and finances into finding serotonin-enhancing drugs, and the new giants of the antidepressant world were born in rapid succession. Prozac was created in 1974. Paxil appeared in 1975, Zoloft in 1977 (the trade names were introduced years later).
In 2003, in Boston, I began treating a 53-year-old woman with advanced pancreatic cancer. Dorothy had no medical problems until she developed an ominous sign known to every cancer specialist: painless jaundice, the sudden yellowing of skin without any associated pinch of discomfort. Painless jaundice can have many causes, but the one that oncologists know best, and fear most, is pancreatic cancer.
In Dorothy's case, the mass in the pancreas turned out to be large and fist-shaped, with malignant extensions that reached backward to grip blood vessels, and a solitary metastasis in the liver. Surgical removal was impossible, chemotherapy the only option.
The suddenness of the diagnosis struck her like an intravenous anaesthetic, instantly numbing everything. As we started chemotherapy in the hospital, she spent her mornings in bed sleeping or staring out of the window at the river below. Most disturbing, I watched as she lapsed into self-neglect. Her previously well-kept hair grew into a matted coil. The clothes that she had worn to the hospital remained unchanged. There were even more troubling signs: tiny abrasions in the skin that were continuously picked at, food left untouched by the bedside table and a gradual withdrawal of eye contact. One morning, I walked into what seemed like a daily emotional flare-up: someone had moved a pillow on the bed, Dorothy had been unable to sleep and it was somehow her son's fault.
This grief, of course, was fully provoked by the somberness of her diagnosis — to not grieve would have been bizarre in these circumstances — but she recognized something troubling in her own reaction and begged for help. I contacted a psychiatrist. With her consent, we prescribed Prozac.
In the first weeks, we waited watchfully, and nothing happened. But when I saw her again in the clinic after a month and a half, there were noticeable changes. Her hair was clean and styled. Her cuts had disappeared, and her skin looked good. Yet she still felt sad beyond measure, she said. She spent her days mostly in bed. The drug certainly affected many of the symptoms of depression, yet had not altered the subjective "feeling" of it. It healed the flaws in her skin but not all the flaws in love.
Any sane reader of this case would argue that a serotonin imbalance was not the initiating cause of Dorothy's depression; it was, quite evidently, the diagnosis of a fatal disease. Should we be searching for a chemical cause and cure when the provocation of grief is so apparent?
Pause for a moment, though, to consider the physiology of a heart attack. A heart attack can be set off by a variety of causes — chronic high blood pressure or pathologically high levels of "bad" cholesterol or smoking. Yet aspirin is an effective treatment of a heart attack regardless of its antecedent cause. Why? Because a heart attack, however it might have been provoked, progresses through a common, final pathway: there must be a clot in a coronary artery that is blocking the flow of blood to the heart. Aspirin helps to inhibit the formation and growth of the clot in the coronary artery. The medicine is clinically effective regardless of what events led to the clot. "Aspirin," as a professor of mine liked to put it, "does not particularly care about your medical history."
Might major depression be like a heart attack, with a central common pathway and with serotonin as its master regulator? There was certainly precedent in the biology of the nervous system for such unifying pathways — for complex mental states triggered by simple chemicals. Fear, for instance, was found to involve a common hormonal cascade, with adrenaline as the main player, even though its initiators (bears, spiders or in-laws) might have little resemblance to one another.
But such a line of inquiry can't tell us whether the absence of serotonin causes depression. For that, we need to know if depressed men and women have measurably lower levels of serotonin or serotonin-metabolites (byproducts of serotonin breakdown), in their brains. In 1975, pathologists performed autopsies on depressed patients to measure serotonin levels. The initial findings were suggestive: depressed patients typically tended to have lower levels of brain serotonin compared with controls. But in 1987, when researchers in Scandinavia performed a similar experiment with newer tools to measure serotonin more accurately, serotonin levels were found to be higher in depressed patients. Further experiments only deepened these contradictions. In some trials, depressed patients were found to have decreased serotonin levels; in others, serotonin was increased; in yet others, there was no difference at all.
What about the converse experiment? In 1994, male subjects at McGill University in Montreal were given a chemical mixture that lowered serotonin. Doctors then measured the fluctuations in the mood of the men as serotonin levels dipped in the blood. Though serotonin was depleted, most of them experienced no significant alterations in their mood.
At first glance, these studies seem to suggest that there is no link between serotonin and depression. But an important fact stands out in the McGill experiment: lowering serotonin does not have any effect on healthy volunteers with no history of depression, but serotonin-lowering has a surprisingly brisk effect on people with a family history of depression. In these subjects, mood dipped sharply when serotonin levels dropped. An earlier version of this experiment, performed at Yale in 1990, generated even more provocative findings. When depressed patients who were already responding to serotonin-enhancing drugs, like Prozac, were fed the serotonin-lowering mixture, they became acutely, often profoundly, depressed. Why would serotonin depletion make such a difference in a patient's mood unless mood in these patients was, indeed, being controlled by serotonin?
Other experiments showed that though depressed patients generally didn't have consistently lower levels of serotonin, suicidal patients often did. Might contemplating suicide be the most extreme form of depression? Or is it a specific subtype of mood disorder that is distinct from all the other forms? And if so, might depression have multiple subtypes — some inherently responsive to treatment with serotonin-enhancing drugs and some inherently resistant?
We may not understand how serotonin-enhancing antidepressants work, but do we know whether they work at all?
In the late 1980s, studies examined the effect of Prozac on depressed subjects. Several of these trials showed Prozac reduced the symptoms of depression when compared with a placebo. Depression is usually assessed using a standardized rating scale of different symptoms. In general, some patients reported clinically meaningful improvements, although the effects were often small and varied from trial to trial. In real-world terms, such a change could be profound: a transformation in anxiety, the lifting of the ache of guilt, an end to the desire to commit suicide. But for other patients, the changes were marginal. Perhaps the most important number that emerged from these trials was the most subjective: 74 percent of the patients reported feeling "much" or "very much" better on antidepressants.
In 1997, a psychologist, Irving Kirsch, currently at the Harvard Medical School, set out to look at the placebo effect in relation to depression. In part, the placebo effect works because the psyche acutely modifies the perception of illness or wellness. Kirsch wondered how powerful this effect might be for drugs that treat depression — where the medical condition itself happens to involve an alteration of the psyche.
To measure this effect, Kirsch combined 38 trials that included patients who had been given antidepressants, placebos or no treatment and then applied mathematical reasoning to estimate how much the placebos contributed to the improvements in mood. The analysis revealed two surprises. First, when Kirsch computed the strength of the placebo effect by combining the trials, he found that 75 percent of an antidepressant's effect could have been obtained merely by taking the placebo. When Kirsch and his collaborators combined the published and unpublished studies of antidepressants (they obtained the unpublished data from the F.D.A. via the Freedom of Information Act), the effects of the antidepressants were even more diluted — in some cases, vanishingly so. Now, the placebo effect swelled to 82 percent (i.e., four-fifths of the benefit might have been obtained by swallowing an inert pill alone). Kirsch came to believe that pharmaceutical companies were exaggerating the benefits of antidepressants by selectively publishing positive studies while suppressing negative ones.
But there are problems in analyzing published and unpublished trials in a "meta-trial." A trial may have been unpublished not just to hide lesser effects but because its quality was poor — because patients were enrolled incorrectly, groups were assigned improperly or the cohort sizes were too small. Patients who are mildly depressed, for example, might have been lumped in with severely depressed patients or with obsessive-compulsives and schizophrenics.
In 2010, researchers revisited Kirsch's analysis using six of the most rigorously conducted studies on antidepressants. The study vindicated Kirsch's conclusions but only to a point. In patients with moderate or mild depression, the benefit of an antidepressant was indeed small, even negligible. But for patients with the most severe forms of depression, the benefit of medications over placebo was substantial. Such patients might have found, as Andrew Solomon did, that they no longer felt "the self slipping out" of their hands. The most severe dips in mood were gradually blunted. Like Dorothy, these patients most likely still experienced sorrow, but they experienced it in ways that were less self-destructive or paralyzing. As Solomon wrote: "The opposite of depression is not happiness, but vitality, and my life, as I write this, is vital."
These slippery, seemingly contradictory studies converge on a surprisingly consistent picture. First, patients with severe depression tend to respond most meaningfully to antidepressants, while patients with moderate or mild depression do not. Second, in a majority of those who do respond, serotonin very likely plays an important role, because depleting serotonin in depressed patients often causes relapses. And third, the brain-as-soup theory — with the depressed brain simply lacking serotonin — was far too naïve.
