Farcical serotonin
How did it all go wrong? The short answer is from the beginning. The germ of the problem was the lack of a germ. The development of iproniazid as a cure for TB was correctly predicated on the identification of Mycobacterium tuberculosis as the germ that caused the disease. Iproniazid was selected from hundreds of other candidate drugs because of its exceptional ability to stop these bacteria from multiplying in a Petri dish or in a mouse. Then it was shown to be effective in patients who were infected with the same bacteria. The scientific logic proceeded securely from a well-validated drug target to a clinically successful medicine. By comparison, the development of iproniazid as a drug for depression was logically back to front. The path started with a clinical effect (elation in TB patients); worked back from that to decide that the drug target was an enzyme that controls the amount of adrenaline in the brain; and then worked back from that to infer that depressed people must have low levels of adrenaline. For TB, the drug was engineered to treat the disease, whereas, for depression, the disease was reverse-engineered to fit the drug.
The development path of Prozac was logically more respectable. The scientists at Lilly started from a target, serotonin, that they thought was a causative factor in depression, based on the high-profile neurotransmitter theories of the time. They worked forward from that starting point to find a drug that hit the target (and only the target); and then they demonstrated that the drug (sometimes) worked in clinical trials. That progression from target to drug to clinical trial is the tried-and-tested path of drug development. It works well if you start from the right place, by picking the right target, a valid target. But if you start from the wrong place, by picking an invalid target, a molecule or a cell or a germ that doesn’t really have anything to do with the disease you’re trying to treat, then following the correct path can still lead you into trouble.
Even as Schildkraut unveiled his new theory of depression he pointed out that there was a hole in it. There was very little evidence in 1965 that patients with depression had abnormal levels of adrenaline or noradrenaline in their brains. The same was true of serotonin when the Lilly scientists started their hunt for the first SSRI about 1975. It was a drug target, fair enough, but not a well-validated one. Medicinal chemistry had advanced to the point that it was increasingly easy to engineer drugs that were designed specifically to disable serotonin reuptake inhibitors, and only serotonin reuptake inhibitors. LY110140, as Prozac was known in the Lilly labs before it became world-famous, had good pharmacology. It hit the target pretty cleanly. But there was never much evidence that serotonin was the right target to be hitting in every patient with depression; and there still isn’t.
We know that serotonin is biologically ancient; it exists in every nervous system, stretching back in evolutionary time from Homo sapiens to the humble worm, Caenorhabditis elegans. Regardless of the species of animal, the number of nerve cells that manufacture and release serotonin is generally very small compared to the number of nerve cells that have serotonin receptors on their surface. In the worm brain, there are only three nerve cells that make serotonin but hundreds of nerve cells that are sensitive to it.38 In the human brain, the serotonin-producing nerve cells are clustered together to form two small nuclei in the brain stem, one of the most primitive parts of the brain, close to its junction with the spinal cord. From this lowly location, about half a million serotonin-producing nerve cells of the human brain send long, branching projections up into the cerebral hemispheres where they make synaptic connections with hundreds of millions of other nerve cells. These evolutionary and anatomical facts have significant implications. They tell us that serotonin is likely to be important for rather basic functions of the nervous system, like the regulation of sleep and eating; otherwise why would the human serotonin system look like a scaled-up version of the worm’s serotonin system? But knowing that serotonin is normally important for brain functions that are disordered in depression is not the same thing as knowing that serotonin deficiency is the cause of depression. To make that claim stand up on its own two feet we need data from depressed patients to show that they have low levels of serotonin in their brains. And this crucial piece of evidence for the serotonin theory of depression has never really materialised, despite decades of searching for it.
I found this out most acutely that day in the outpatient clinic at the Maudsley Hospital when I was assuring a patient that SSRIs would rebalance the level of serotonin in his brain. “How do you know that about me,” he asked, “how do you know that the level of serotonin is imbalanced in my brain?” We both immediately knew that I didn’t have an answer to that question. I didn’t even have a clue about how to find out the answer. After this silent moment of truth, we politely carried on in the customary way. He left with a prescription for an SSRI and an appointment to return in six weeks to tell me if it had made any difference. He left me feeling like a total fraud. For the first time in my medical career, I had seen myself as if performing the farcical role of one of Molière’s ludicrous “leeches”, a vacuous 17th-century physician, telling patients they needed to be bled because of their superfluous sanguinity, without really knowing how much blood they had or how much they needed.
Bereft of biomarkers
Which brings me finally to what I now think is the simplest way of saying why it all went wrong after Prozac: no biomarkers.
In most areas of medicine, doctors are using biomarkers all the time. A biomarker is just a measurement of a biological function or a biochemical in patients. Haemoglobin is a bread-and-butter biomarker, easily measurable by a blood test, that can be used to diagnose anaemia, or too few red cells in the blood. Haemoglobin can also be used to predict the response of an anaemic patient to treatment with a blood transfusion, or to identify the much rarer patients with too many red cells in circulation, who might actually benefit from a bleeding operation as prescribed by one of Molière’s physicians. Haemoglobin is thus, in technical parlance, both a diagnostic and a predictive biomarker. Glucose is another familiar example of a biomarker that is both diagnostic, of diabetes mellitus, and predictive, of therapeutic response to insulin. There are already hundreds of thousands of biomarkers, and they are rapidly growing in number and sophistication, in all areas of medicine . . . except psychiatry, which doesn’t currently have a single blood test or biomarker to its name.39
In a rational universe, the use of SSRIs, and the serotonin theory of depression, would be informed and justified by serotonin biomarkers. When a patient came to see me for advice about how to treat his depression, I would measure the serotonin level in his brain and, if this level was low, I would recommend a drug that was likely to increase serotonin. We could repeat the brain biomarker measurement a few weeks after start of treatment, to check that serotonin levels were returning to normal. Serotonin biomarkers would allow us to use SSRIs without wishful thinking or farcical hand-waving, and much to the benefit of patients. But serotonin biomarkers have never materialised in clinical practice and they are very challenging to measure even in a specialised research study.
The fundamental difficulty in measuring serotonin biomarkers relates to the anatomy of the serotonin system. There aren’t very many serotonin-manufacturing nerve cells in the human brain and they’re mostly concentrated in a few small clusters in the brain stem. The only conceivable way of measuring the level of serotonin in these cells, in a living human, is by brain scanning or neuroimaging. And in practice it is very difficult to make any kind of image of such a small and inaccessible part of the brain. Some studies have used special scanners to measure levels of serotonin transporters in patients with depression.40 But the requisite technology is expensive, difficult to use outside a few specialist centres, and involves the patient receiving a small but significant dose of a radioactively labelled drug. It could never be used as a biomarker in everyday clinical practice and it has not been much used as such in research studies of depression and serotonin.
The only other optio
ns are to measure serotonin and related molecules in the blood; or in the cerebrospinal fluid (CSF), the watery liquid that flows through the internal chambers or ventricles of the brain. Both options have been explored in research studies but not pursued into practice. Blood biomarkers of serotonin have not been reliably diagnostic of depression, or predictive of response to SSRIs, and are probably not very representative of serotonin levels in the brain. CSF biomarkers are more likely to be representative of brain serotonin levels than blood biomarkers. But taking a CSF sample for molecular analysis requires a lumbar puncture or spinal tap: inserting a long needle between two vertebrae at the base of the spine, and withdrawing a couple of teaspoons of fluid. The extra diagnostic gain from CSF serotonin biomarkers has not justified the extra pain of lumbar punctures.
So it is not for want of trying that we don’t have biomarkers to guide serotonin-tweaking drug treatments for depression. But we don’t. And in the absence of biomarkers, we will never have a straight answer for patients about why they should take SSRIs. We will continue to proceed by trial and error, trying one drug then another if the first one doesn’t work. Perhaps most regrettably, we will be encouraged to behave as if all depression were the same. If we can’t tell the difference between the depressed patients with high serotonin levels and those with low serotonin levels, then we will often assume, as I did in the Maudsley Hospital outpatient clinic, that they must all have low serotonin levels, they must all be the same, to justify prescribing the same first-line treatment for everyone with depression.
Yet when we say they are all the same, or when we don’t say it out loud but we act as if it was the case, that all depressed people are depressed for the same reason and they are all likely to benefit from the same treatment, we should stop and think about what we mean by the word “they”. In this context, it represents about 10% of the global population at any one time, or about 25% of everyone over the course of their lifetimes, or at least one member of all the families on the planet. I would dare to say that none of us will live our lives untouched by depression, directly or indirectly. So there isn’t really much difference between “them” and “us”, when it comes to depression, although the culture of stigmatisation would have you believe otherwise. And, for me at least, it beggars belief that such a huge chunk of humanity should suffer from the oscillations of a single unmeasurable molecule in the brain. The serotonin theory in that sense is as unsatisfactory as the Freudian theory of unquantifiable libido or the Hippocratic theory of non-existent black bile.
• • •
In short, depression after Descartes is in a sorry state. According to the prevailing dualist orthodoxy that he bequeathed us, it is officially a disorder of the mind, and as such it may be exacerbated by stigmatisation, or alleviated by psychotherapy. But it is also treated unofficially as if it was a disorder of the brain, using drugs for which we lack a rationale beyond the reach of satire. We deal with depression as if it is not entirely of the mind but not truly part of the brain or body either. We don’t agree with each other about how to deal with it better - there is grumbling contention, a culture war, a bit of name calling, between psychologically minded advocates of a more “brainless” approach, on the one hand, and neuroscience-driven advocates of a more “mindless” approach, on the other. Meanwhile, we have had no major new treatments for a generation and the limitations of the existing drugs and talking therapies are obvious. Despite increasing access to psychotherapy, despite increasing numbers of prescriptions for SSRIs at decreasing cost per pill, depression is still expected to be the single biggest cause of disability in the world by 2030. It is not cancer, or heart disease, or rheumatoid arthritis, or TB, or any other physical disease, that accounts for economic costs in the order of 3% of GDP in rich countries. It is mental health disorders, principally depression. And we don’t really know what to say or do about it.
It’s time to turn the page.
Chapter 5
HOW?
Extraordinary claims demand extraordinary evidence
Around the time of Prozac’s launch, circa 1990, as the wave of therapeutic excitement triggered by the accidental antidepressant discoveries of the 1950s was peaking, a few papers were obscurely published. They had titles like “Stress and immunity: an integrated view of relationships between the brain and the immune system” (1989);41 “The macrophage theory of depression” (1991);5 “Evidence for an immune response in major depression” (1995).6 These and related papers were necessarily published in obscure journals because their scientific hypothesis could hardly have been more transgressive, more ideologically scandalous - in a word, flakier. They proposed that mood states were somehow related to the activity of white blood cells, that the mind was connected to the body across the Cartesian divide. In the scientific climate of the time, this idea was not even wrong. It was worse than that. It was tantamount to proposing that mood states were related to the flow of black bile and other occult humours. And, quite properly, for many years this theory was either ignored or attended to only with a high degree of scepticism by most other scientists.
It is often difficult to pinpoint historically when a new scientific theory breaks through. We all stand on each other’s shoulders, almost all new ideas are incrementally derived from older ideas, and the book of knowledge grows gradually, as a result of the collective efforts of many individual scientists working on the same page. It is even more difficult to recognise a breakthrough at the time it is happening, rather than retrospectively, because, by definition, a breakthrough breaks things. It must disrupt or unsettle or undermine or challenge some prior certainty. So, at the moment of its insurgency, a scientific breakthrough will be resisted, denied, obscured or ridiculed by all right-minded supporters of the status quo.
Those early papers in neuro-immunology or immunopsychiatry now look like a breakthrough to me. Their shared idea - that mood and inflammation are linked - was a scientific reformulation of what is common knowledge among patients, if not among their physicians. We all know that mood disorders are closely associated with physical disorders. All of us have had occasional experiences of fatigue, social withdrawal, low mood and other depressive symptoms after a physical injury, like a bone fracture or dental surgery, a chest infection or a vaccination. It seems obvious to many nonmedical people that physical and mental health are closely linked. The innovative idea of immuno-psychiatry is that this association is explained by the immune system. And to test this idea, scientists conducted the first experiments to measure biomarkers of inflammation - white blood cells and cytokines - in depressed patients.
This was unprecedented: it was the first time that we began to apply the power and precision of modern immunology to help us understand human behaviour and depression. And for at least 15 years this seminal endeavour was conventionally regarded as beyond the pale. In 2012, when I caught up with the story, when I first began to see how there might be something in it, that immunological mechanisms might cause depression, and that anti-inflammatory drugs might even be a new type of anti-depressant, I consulted senior colleagues, as you do.
“I’d always thought you were more sensible than that,” said the Regius Professor of Physic in the University of Cambridge (who was sort of joking). “If you’d come to me with this idea five years ago, I’d have thought you were crazy; but now I’m not so sure,” said the Senior Vice President of R&D in GlaxoSmithKline (who wasn’t joking).
You won’t be surprised that I think the core assumption of Cartesian dualism - the conventional idea that mind and body are distinct domains - lies deep beneath this scepticism. But if you scratch the surface of their resistance to neuroimmunology or immuno-psychiatry, sophisticated scientists these days won’t invoke Descartes by name (they mostly think philosophy is irrelevant to the day job). Instead they will ask about evidence, causality and mechanism. They want to get to the crux of the matter.
Scientists want to be convinced that there really is a causal relationship between inflammation and
depression. Then they want to know how and why.
How, exactly, step by step, can inflammatory changes in the body’s immune system cause changes in the way the brain works so as to make people feel depressed?
Why is the depressed patient inflamed in the first place? And why should the body’s inflammatory response, which is supposed to be on our side, which has evolved to help us win the battle against disease, be causing us to feel depressed?
The Inflamed Mind Page 10