I'm going to focus on how antidepressants work and from that perspective, what do we make of all the different categories that are created for antidepressant medications.

This highlights the fact that the monoamine systems in the brain have overlapping but also slightly different profiles, and most of the data are not from people with depression. They are from animal studies. Some of these studies in healthy people look at the effects of drugs that selectively increase norepinephrine, dopamine, or serotonin. One could conclude that perhaps antidepressants work slightly differently in different people.

However, the clinical data tend not to support this well. I happen to think this is accurate. However, I think the way that we collect clinical data is actually making it difficult for us to be able to identify some of the extreme categories in terms of the effects.

Another point is that even though it's likely that norepinephrine affects energy and interest, for instance, more than serotonin does, I think it's a fatal flaw to make the assumption then that too little norepinephrine causes all conditions that have difficulties with energy or interest. In fact, I think it's clear that energy and interest disorders relate to the part of the brain that norepinephrine interacts with, and it's problems in the part of the brain, probably more so than a deficiency or dysfunction of norepinephrine as an example, that causes energy and concentration disorders in humans. That'll be a central theme of the latter part of the presentation.

This is the alphabet soup part. I think that if you take antidepressants that we currently have, and you group them, you basically come up with monoamine oxidase inhibitors (MAOIs), tricyclic antidepressants (TCAs), selective serotonin reuptake inhibitors (SSRIs), serotonin and norepinephrine reuptake inhibitors (SNRIs), norepinephrine reuptake inhibitors (NRIs), receptor agonists and antagonists (RAs), and monoamine releasing agents (MRAs). We don't really know how antidepressants work and so why even go through this process of categorizing them in this way.

Plus, you treat some patients with one and some with another and it doesn't really matter. And the focus of the rest of the presentation is, does it really matter; do these categories have any significance and if so, how would you use that clinically? I'm going to talk about more of the theory and a lot of the rationale for why we divide drugs up like this. The detail in regards to some of the clinical effects is really where "the rubber meets the road"; it's an understanding whether or not in a real patient, in a clinical setting, you actually see the kinds of differences that I'm going to predict that we might see, based on some of the pharmacology and the basic science of the medications.

Here are most of the common antidepressants that are available, ranked based on the potency and blocking reuptakes of either norepinephrine or serotonin (5-HT). And on the far right, is the ratio of the potency or affinity for binding to the norepinephrine transporter relative to the serotonin transporter. The reason for showing this is 2-fold.

One is that all existing antidepressants, even some of the ones that are not listed here because they don't block reuptake like mirtazapine, share potent effects on increasing brain levels of either norepinephrine, serotonin, or both. And even though sometimes you can say, drug "X" does this, this, and this; drug "Y" does that, that, and that. Ultimately, the strongest evidence suggests that the core proximal mechanism of action of antidepressants that we have today is, in fact, raising levels of norepinephrine and/or serotonin. Other effects appear not to be necessary because once we eliminate those effects in specific drugs, or at least in the drugs developed without those effects, continue to work.

So you can create selective drugs that only block norepinephrine reuptake, and the 2 examples at the top, reboxetine and atomoxetine, are such medications. Atomoxetine is now available for attention deficit hyperactivity disorder (ADHD) in the United States; reboxetine is available in Europe as an antidepressant. Those are the 2 most potent and selective NRIs. Small numbers mean high potency and large numbers mean low potency; weak kinds of effects.

Reboxetine and atomoxetine have the highest affinity for blocking norepinephrine reuptake, bupropion, at the bottom, has virtually no effect, and escitalopram is relatively weak. You switch from norepinephrine to serotonin, what you see is that escitalopram is probably one of the more potent and certainly the most selective. And as you go up the list, you get increasing fewer effects on serotonin in terms of the potency or the affinity of blocking the reuptake.

If you look at the ratio, reboxetine and atomoxetine, desipramine and amitriptyline tend to have relatively potent effects. As you go towards the middle, you get dual action, and the drugs at the bottom, with the exception of bupropion, have very potent and selective effects towards serotonin with very little effects on norepinephrine. I would argue that in the clinic you can see subtle differences between these drugs and that you would not use them the same way. The dual action drugs have an overlapping effect that crosses both categories.

From this point of view you can say, we know that this is true, that these drugs bind in that potency and have that affinity. A lot of people will ignore ratio. If you look at the ratio, it doesn't mean anything to them. But I think the ratio means something. It says that you know for drugs like reboxetine and atomoxetine, it's very difficult to get to the point where you're blocking serotonin reuptake unless you get into toxic doses. Likewise, for escitalopram and citalopram it is very difficult to get an effect on norepinephrine unless you go up into toxic doses. Paroxetine, which has a high affinity, 45 (low number means high affinity), has a fair amount of potency there, but if you look at the profile, it's 450-fold more potent on serotonin. So the kind of doses that you're likely to have to use clinically before you get norepinephrine reuptake inhibition, even though it has high affinity, is likely to be above or at least at the very top end of the clinical dose range. I think from this you could fairly argue that at least the extreme drugs are going to have relatively selective effects in regards to these 2 neurotransmitter systems.

Other effects are listed here: alpha 1, alpha-2 receptor, beta, histamine, and muscarinic, cholinergic, and dopamine receptor bindings. And what you see is that most of the drugs that bind to these are known for the side effects that are related to those. And if you eliminate those effects, the drugs continue to work. There was a time when people argued that anticholinergic or antihistamine effects were important antidepressant actions. I think that you can see here that you can basically eliminate those effects, and still get very effective, potent antidepressants. Those effects aren't necessary for antidepressant action. Whether or not they add something is another question, but certainly they aren't necessary to having the effect.

The other point I'd make is that if you look at antidepressants, it is clear that within the first few minutes, the first hour, you get increases in the transmitter that the drug works on. This is a slide from fluvoxamine (Luvox), and what you see is that fluvoxamine increases levels of serotonin very quickly. These fraction numbers are within the first few minutes, and norepinephrine levels don't change.

There are data in laboratory animals showing that selective drugs have selective effects, drugs like mirtazapine, which increase norepinephrine and serotonin indirectly by blocking out the 2 receptors; also within the first few minutes increased levels of norepinephrine and serotonin. So the first time you take that first pill within the first hour, you've got 150% to 200% increase in levels of norepinephrine and serotonin in the brain. It's true that if you wait longer, the amount does increase a bit more; however, it's not fair to say that you don't get an effect early on. In fact, you get a relatively potent effect; 150% to 200% increase in levels of these transmitters.

If you look at bupropion, you get increases in dopamine and norepinephrine in the brain with microdialysis studies. Bupropion increases dopamine and norepinephrine; selective NRIs increase norepinephrine, they also increase dopamine a little bit too indirectly, and SSRIs increase serotonin.

This is looking at a human and looking at a positron emission tomography (PET) scan at baseline and then a PET scan later, on paroxetine. For those of you who aren't used to looking at these kinds of figures, the main thing to focus on is the 2 bright areas in the middle scan, up towards the middle. You see these bright areas, and that indicates a high amount of binding to the serotonin transporter. After you give paroxetine, you displace that so it gets dimmer. And what you can see there is that in a human, you diminish the amount of binding to the transporter, indicating that paroxetine really does bind to the serotonin transporter in the brain.

If you look at bupropion and if you look at binding to the dopamine transporter (these are relatively recent data about to come out in Biological Psychiatry), you see a very similar thing, and pay attention to the red colors that indicate a lot of binding.

Because you're giving a drug that displaces the radiolabeled ligand, you're getting a duller and duller signal over time. And so, this shows that with bupropion at clinical doses, you block dopamine reuptake in the brain.

The reason why you don't get drug abuse potential with bupropion is shown in this slide from Volkow. If you look at cocaine, you get very potent, very rapid binding (blue line), and bupropion is very slow and doesn't actually get to the same point. So drugs that get onto the reuptake carrier and block it slowly have a different profile of effects even though it's working on the same system.

From these data -- and I haven't really surveyed the entire field, it's just kind of a sampling -- it's very clear that the drugs that we have block either norepinephrine reuptake, serotonin reuptake, or both, and that those are the most potent pharmacological effects.

How do we know that that really relates to how they work? Here's a second piece of data that highlights the fact that these drugs are working by changing those neurotransmitters. And that's some of the work that I've done and colleagues at Yale, Arizona, and now in Ohio, depleting serotonin and/or dopamine in people who have taken antidepressants and are being treated. And by giving an inhibitor of the synthesis of norepinephrine and dopamine, alpha-methylparatyrosine (AMPT), we can block the production of norepinephrine and dopamine. There is an extensive, 30-year history of work that's been done with AMPT in people and in laboratory animals, all of it highly consistently showing that you could lower norepinephrine and dopamine within 12 hours in an animal and in a person with administration of these drugs that block the synthesis.

Likewise, you can lower serotonin in the brain by depleting the precursor amino acid, tryptophan.

Tryptophan can be depleted by giving an imbalanced mixture of amino acids, shown here in the white powder.

These are the amino acids. So by giving this mixture of amino acids, you can fool the body into using up its tryptophan, and tryptophan is important for making serotonin; in fact, it's the only way that you can make it. And if you leave tryptophan out but you give these amino acids, which is about 100 grams total, tryptophan is a relatively small percentage. If there's no tryptophan there, your body wants to use it, kind of like your body is using what you just ate to make proteins and other components. But with these amino acids, your body starts cranking out the proteins and there's no tryptophan, and the only place to get the tryptophan is to suck it out of blood. And so tryptophan levels plummet by about 90% in 5 hours.

If you look in the brain, and this is from a PET scan study that was done in McGill University, you see serotonin synthesis rates in people before and 5 hours after taking an imbalanced mixture of amino of acids, and you do see lowering of serotonin in the brain.

The procedures that we began to apply to people with depression, to understand antidepressant action, are changing levels of serotonin in the brain.

These are data from about 15 years of studies in various groups of people with depression and nothing else. These are unipolar patients with major depression. Some were treated with SSRIs, which include all of the SSRIs except citalopram, which wasn't available when we did these studies. NRIs are in the middle, which is mostly desipramine and a few patients on bupropion, and on the right are healthy people and patients with depression who have not been treated and are currently depressed and in an episode.

If you look down at the bottom, the most common effect of depleting neurotransmitters in people is no effect. Most people are unchanged. So if you look in the red, no symptoms. You can deplete norepinephrine and dopamine, or serotonin, or both in healthy people and you don't induce clinical depression. You can deplete those neurotransmitters on the far right in patients who are depressed before treatment and they don't become worse; even if they're only mildly depressed, even if they're severely depressed, they don't worsen.

It's only when you look at people who are on antidepressants and raise levels of a certain neurotransmitter that you begin to see effects, and there the effects are relatively selective; not entirely but relatively. Patients taking desipramine or bupropion, basically you can deplete serotonin and the vast majority of those patients are walking around as if nothing is going on. These people are continuing to take their antidepressants, they go through this depletion testing, and nothing happens. However, if you deplete norepinephrine and dopamine, their symptoms come back. Then they go away again as the effects of their depletion wear off.

If you look at people treated with SSRIs, what you see is the depletion of serotonin causing an effect in about 60% to 80%, but depletion of norepinephrine and dopamine doesn't. This argues that whether or not you see it clinically, selectivity does matter; that these systems, even though we know there's a lot of interaction in the brain, seem to behave partially independently in humans taking antidepressants for depression. You can undo the effect of the drug only by depleting the transmitter that that drug increases.

I think those 2 pieces of information -- first that the drugs are selective and that selective drugs work; secondly that you can deplete the neurotransmitter and undo the therapeutic effect of the drug but not cause the symptoms by depleting a transmitter that that drug doesn't increase -- suggest that these pathways are doing something a little bit differently than each other.

When we looked at the symptoms that emerged in people who had depressive recurrences during these depletions, what we found is a profile that is shown here. Norepinephrine and dopamine depletion led to a slightly more robust effect at causing an increase. High bars mean more symptoms, a worsening of concentration, interests, energy, and retardation. Serotonin depletion led to a greater amount of anxiety than norepinephrine/dopamine depletion in patients who were taking antidepressants.

What you see here is a profile that I suspect for most of you is going to be reminiscent of what you see in the clinic. In all of the patients, all those symptoms changed and the mood symptoms changed comparably for either type of depletion. So it's clear that there's an overlapping area and it's also suggestive that in some people there may be a slightly greater profile going in the direction that we would have predicted at the very beginning, looking at that Venn diagram of how norepinephrine, serotonin, and dopamine affect the brain.

From these studies you can conclude that antidepressants probably do have more than 1 mechanism of action, that matters; norepinephrine and serotonin at least are involved and probably dopamine as well. However, I think that norepinephrine and dopamine depletion are unlikely for dysfunction, unlikely to be the cause of depression because if they were, we should make depressed people who are untreated worse and we should make healthy people depressed, and we don't. We also should see symptoms in someone taking an SSRI if you deplete norepinephrine and you don't, and you should see symptoms in someone taking a norepinephrine drug by depleting serotonin and you don't.

It's hard to reconcile that with the model, where depression has to do with a deficiency or dysfunction of either of these neurotransmitters. We're used to looking at the nervous system, the brain in this way. We think about neurons in very simplistic terms. In fact, they're much more complicated than that. Neurons are more alive and they're much more branched.

This is actually my arm, and the point is that if you had a rash on the arm, got a corticosteroid cream and put it on the rash, and the rash got better after a couple of weeks, you wouldn't automatically conclude that rashes were due to corticosteroid cream deficiency or due to a dysfunction of cortisol in your body. In fact, cortisol deficiencies cause other symptoms, but they don't cause rashes. Cortisol deficiency can make you more likely to react if you're predisposed. Rashes are caused by a lot of different things, and corticosteroids interact with inflammatory processes to reduce them.

I would speculate not that antidepressants are working through steroids, but that antidepressants are working in a steroid-like manner in the sense that they induce a time-dependent series of effects through receptors that are specific, and that ultimately leads to improvement in depression.

It's important to understand that behavior is really a process that is of a much higher order of complexity than neurotransmitter levels and that behavior results from cellular interactions between brain areas.

Psychiatry was waylaid about 40 or 50 years ago by neurologists. They discovered that people with Parkinson's disease had a deficiency of dopamine and that giving a precursor improved the disease. There would be studies and presentations where they bring in the guy with Parkinson's, he walks across the stage, they take him backstage and they give him some levodopa (L-dopa), and he comes back just moving around. This was in the late 1950s, early 1960s. In fact, that was the first brain disorder where a cause, or at least a proximal cause, had been identified.

If that happens in Parkinson's disease then it must happen in schizophrenia, depression, bipolar disorder, and if our drugs increased levels of norepinephrine and serotonin, depression must be a dysfunction of these systems.

I think we got hung up on neurotransmitters. In fact, most psychiatrists, if you asked them to take a brain and identify the hippocampus, the amygdala, the frontal cortex, prefrontal cortex, or cingulate, probably would be wrong 50% of the time. While transmitters are important, in regards to behavior, the most important aspect of brain function is really the circuitry that modulates emotions and behaviors, at least in people with mental disorders. The circuitry is actually R circuits that are made up of neurons that do not involve norepinephrine and dopamine or serotonin. Norepinephrine, dopamine, and serotonin modulate excitatory amino acid circuits, GABAergic circuits in discrete parts of the brain, and when you modulate the functioning of those circuits, you begin to change behavior.

Where we should be looking in the pathophysiology of these conditions is in those brain areas, in those cells. And we should be looking for cellular dysfunction in the brain regions that are modulated by norepinephrine, serotonin, and dopamine, which include the frontal cortex, basal ganglia, hippocampus, and amygdala. That's where all the brain imaging is done. That's what people are looking at, and the reason that they're looking at the anterior cingulated and the frontal cortex is not because serotonin is there; it's because these circuits are involved in modulating emotions, sleep, sexuality; all of the functions that we see disordered in mental disorders.

The immediate effects, the proximal effects of antidepressants were only the first step in a cascade. And that cascade takes you all the way inside the cell, so that a neurotransmitter interacting with a receptor through a G-protein, which is a signaling protein, and eventually second and third messenger systems, begins to alter the production of a variety of compounds inside the cells.

Where are these cells? These cells are not norepinephrine, serotonin, or dopamine neurons; the cells that are being modulated by norepinephrine, dopamine, and serotonin to produce the proteins are cells in the frontal cortex, hippocampus, amygdala and other areas of the limbic system.

The effect of serotonin on these cells begins to have a hormone-like or fertilizer-like effect; antidepressants begin to modulate activity in a fertilizer-like fashion and that may account for a time delay in antidepressant action.

Instead of a simple neuron this is closer to what a real neuron looks like; a lot of input and a lot of output, and those outputs, the apical dendrites, can actually increase or decrease.

Looking at the varicosities at the end of a neuron, this is part of the work that ultimately Kandel got the Nobel Prize for. He showed that learning is associated with structural physical changes in nerve cells. It's the increase after sensitization, after learning, in simple and in more complex animals, this increase in arborization and the formation of new connections by neurons, which happens in a few weeks, that underlies learning.

If you look at the neurons, they're little hairy projections at the end of them. If you look over time (to the right), learning is associated with hairiness; the neuron grows and forms connections; that's a normal process.

However, it's become clear as well, that stress reduces that. Stress prunes all of this kind of increase in growth, and the areas of the brain that are most sensitive to this are areas that are involved in emotion regulation, the limbic system.

The limbic neurons are very stress sensitive and they can shrink and lose connections with stress, and that's a normal process. If you take the stress away, the neuron goes right back. That's assuming a normal system. And for reasons that are outlined here, genetic factors, acquired factors, this process can go awry.

So one of the newer theories about what's wrong in depression is that it's this cellular process in neurons that make up the circuits in the frontal cortex, amygdala, and hippocampus, the limbic system, that is actually at fault in depression for a variety of genetic and acquired reasons. Anything that can make those neurons function less well is going to make them bounce into and out of the effects that stress causes less easily. In fact, if those neurons are really not working very well, they might get stuck and not bounce back, even though the stress is gone.

Ultimately this model is a slightly more complex way of explaining what might be going on in depression, why depression is associated with stressful life events, and why is it sometimes that psychotherapeutic techniques, talking therapies or other kinds of experiences that people have, spiritual experiences, and holistic experiences that lead them to feel a bit better and less stressed, might actually be part of the beneficial effect? You'd also predict that if this process is a normal process, if you reduce stress, that the neurons might be capable of bouncing back.

Ultimately, I think this model helps. The most important thing I'd point out in relation to antidepressants is that norepinephrine and serotonin tend to reverse the stress effects in laboratory animals; they cause the neurons to regrow in a fertilizer-like fashion, interacting with neurotrophic factors.

Summing up, I think that even though norepinephrine and serotonin seem to have divergent effects, there may be a core set of effects that they may have in common and that projection to other parts of the brain may account for some of the differences that we see between antidepressants. I think the challenge is in clinical studies, do you really see these kinds of effects? Because based on depletion work and the pharmacology of antidepressants you'd predict that drugs that are selected for norepinephrine and serotonin should have a different profile.

If you wanted to identify differences between antidepressants in regard to noradrenergic effects and serotonergic effects and dopaminergic effects, you'd have to use rating scales that are different than the Hamilton Depression Rating Scale. And you'd have to select patients who are mostly lethargic, or anxious and irritated, or impulsive/compulsive, and then look to see whether the drugs have differences.

Based on this, I think it's clear that the pharmacology does matter and ultimately the question is: are we going to see those effects in the clinic?