Dr. Doghramji left us with the question of whether treatment can reverse some of the abnormalities in insomnia, and that's pretty much the theme that I would like to pick up; to talk about the treatments that currently exist. But I'd also like to go a little bit beyond that to tell you a bit about some of the treatments that are in the pipeline.

So we're going to review very briefly a couple of concepts related to behavioral and psychological treatment strategies; we'll talk about the currently available agents, which are essentially the benzodiazepine receptor agonists; and we'll look at some of the other drugs used to treat insomnia as well. And then I'll go into some of the treatments that are under development that we have a bit of data on, and that we'll be hearing more about in the future. It's important to note that when we're talking about these other pharmacologic treatments, besides benzodiazepine receptor agonists, we're talking about a series of drugs that are not approved for the indication of insomnia. They have generally received much less validation through research studies, so these are clearly off-label indications.

First of all, before we get into the pharmacologic treatments, I'd like to talk a bit about behavioral and psychological treatment strategies.

This is my own definition of behavioral treatments for insomnia; I think of them as a diverse set of behavioral prescriptions designed to improve nocturnal sleep quality. There's 2 parts of that; the first part is that these are a diverse set of behavioral prescriptions. These are things that we work with our patients to accomplish but they are not things that we can give to our patient; the patient has to be a willing and active collaborator in this because it involves changes in behavior. And as we all know, change in behavior is easy to talk about and very difficult to do, like so many of the major medical conditions facing society today, from overweight to smoking, alcohol abuse, and insomnia. They all involve major behavioral components and sometimes can be hard to change. I think insomnia is one of the easier sets of behaviors to change, and we should keep in mind that behavioral treatment should always be in our treatment armamentarium.

So these are treatments of the patient, by the patient, and for the patient. Compliance is an issue because we can, again, recommend and prescribe to patients what they should do, but it depends on a change in their own behavior. What I like to tell my patients is that it's important to keep in mind that you obviously can't make yourself go to sleep because if people could make themselves go to sleep, they wouldn't be seeing me. So the notion is that sleep itself is an involuntary behavior and what you have to do is to work on voluntary changes to the part you can control, which is your waking behavior, as a way of influencing sleep. All of these things are designed to improve the quality of nocturnal sleep, and studies that I will reference show that they do work and they have a number of common elements as well, and those regular elements include regularizing sleep timing and increasing predictability of sleep.

Just a brief review to show you that these treatments do in fact work. There have been a number of meta-analyses. These are some results from Smith and colleagues who had compared behavioral and pharmacologic treatments for insomnia. Basically what you see is that both sorts of treatments resulted in improvements in wake after sleep onset and total sleep time from pre- to posttreatment. So it's important to keep in mind that this is another efficacious set of treatments that we have available.

I personally like to focus on a couple of simple elements with my patients. I always show them the Borbely 2-process model of sleep regulation to explain that. There's the homeostatic sleep drive that's just a measure of how long you've been awake and it makes sense to patients that if you've been awake for a long time, you get sleepier. But that's not the whole story because if it were, people could go to sleep anytime they wanted and wake up anytime they wanted as long as they got their 16 hours of wakefulness. People are diurnal creatures; they are supposed to be asleep in about an hour, and that's because of the circadian clock. So we have low sleep drive during the day, high sleep drive at night; those 2 things work together.

So the behavioral treatments relate to this model. Restricting time in bed increases homeostatic sleep drive. Establishing a regular wake-up time helps to set the biological clock, so we get light, activity, and social interaction at the circadian system's most vulnerable phase to resetting, which is first thing in the morning, as soon as you wake up. Go to bed only when you're sleepy, which is appealing to the homeostatic mechanism. Stay in bed only if you're actually asleep, which doesn't really relate to the 2-process model; it gets more to conditioning factors, but I think it's worth throwing in there. And then talk about practices that help and hurt sleep.

Those are usually called sleep hygiene factors, which include things such as getting exercise, following regular daily routines, making sure that comorbid medical problems are treated, having a good sleep environment, etc. Conversely, stay away from things that hurt your sleep, such as alcohol and caffeine use and handling worries; there are ways to do that that I'm not going to get into. But it pays off to attend to those factors that help and hurt sleep.

We're going to move on and talk about benzodiazepine receptor agonists.

In a summary from the National Institutes of Health (NIH) State of the Science Conference that was held recently, the evidence supports the efficacy of both cognitive behavioral treatments for insomnia and benzodiazepine receptor agonists. And the second statement, which I think is very important to people in the community and to our patients, is that there's very little evidence to support the efficacy of other treatments despite their widespread use. What are we talking about? We're talking about alcohol; we're talking about over-the-counter things such as antihistamines; we're even talking about antidepressants and we'll get to that.

Also there's a mismatch between the fact that insomnia very often is a chronic condition and almost all of our research evidence regarding its treatment comes from short-term treatment studies. We have a long-term disorder, short-term treatments; there's a mismatch there. So the panel called for more long-term studies to investigate the real efficacy of pharmacologic and behavioral treatments.

It's interesting to note that despite the fact that insomnia is chronic, only 1 medication has been approved for use without a specified time limit — maybe it's something special about eszopiclone. Personally, I don't think that's the case; this drug has a common mechanism of action seen with other drugs. The fact is that other drugs have not been assessed; when that happens, I think that the recommendations are likely to change.

The final thing is tolerance and abuse, which often are concerns to my patients and probably yours. Actually there are not a lot of data to suggest that these are widespread problems in the population that receives hypnotic medications for insomnia.

Let's turn briefly to some receptor pharmacology. Benzodiazepine receptor agonists work by enhancing the activity of gamma-aminobutyric acid (GABA). There are 2 broad sets of GABA receptors in the central nervous system (CNS) — GABA_B receptors, which basically have nothing to do with sleep and we won't say anything more about, and the GABA_A receptor, which really is where the action is. The GABA_A receptor that you see in an artist's depiction here contains 5 protein subunits that have Greek letters, and in the middle is a chloride ion channel. There are recognition sites in the GABA receptor for GABA itself, which is at the junction of the alpha and beta subunits, and there's a recognition site for benzodiazepine and related drugs that's at the intersection of the alpha and gamma subunits.

There are lots of different types of GABA_A receptors in the CNS as well; the way they differ is that they have different arrangements of Greek letters contributing to the receptor complex. The ones that we're concerned about for insomnia and for hypnotic efficacy generally have an alpha subtype-1 protein subunit associated with them. Alpha-1 selectivity means that it's selective for those GABA receptors that have an alpha-1 protein subunit.

In any case, when you get GABA binding, you get chloride-ion influx and chloride-ion influx hyperpolarizes a neuron; in general, a hyperpolarized neuron is a less active neuron. And benzodiazepine and similar drugs increase the affinity of this receptor for GABA, and they actually increase the frequency of the chloride-ion channel openings, resulting in even more hyperpolarization. So all benzodiazepine receptor agonist drugs, whether they're true benzodiazepines or the so-called nonbenzodiazepines, have that common mechanism of action.

The way they differ from each other mainly is in terms of pharmacokinetics. The way these agents differ is in terms of their half-life, ranging from the very short to the very long; some of them have active metabolites. This particular variation of the table also has the cost of 30 doses, and I think this is kind of interesting because there's a 10-fold difference from the cheapest to the most expensive. Who even knows what quazepam is? Interestingly, that's the most expensive one.

This is a study of eszopiclone, 6 months of nightly use. The dashed and dotted and solid lines are whether you're looking at the mean or median. Basically, the story here is that patients showed a rapid improvement in wakefulness after sleep onset that was generally sustained over the 6 months of nightly treatment. The placebo group showed a small improvement over the same interval. So this challenges the notion that these drugs only should be used short term. Patients generally seemed to do fairly well in that study.

One of the interesting aspects of that study was that they also asked people how alert they felt during the daytime; this is not an objective measure, but it's a self-rating of how alert they felt during the day. And you see a very similar pattern, that the people treated with the active drug tended to improve quickly and stayed improved over the duration of the study compared with people similarly treated with placebo.

There are other benzodiazepine receptor agonist drugs in development and under study. Indiplon is one of those. This is, structurally speaking, not a benzodiazepine drug but it acts at the same GABA receptors that have an alpha-1 subunit. It's being developed in 2 formulations, an immediate- and a modified-release. These drugs have the same half-life because it's the same drug, but one is released both immediately and more slowly during the night for extended coverage during the middle of the night. Emerging data from studies of this drug show that it, too, seems to have efficacy for longer than the usual treatment periods that we consider.

This is the immediate-release preparation and time-to-sleep onset, looking at 10 mg and 20 mg of the drug compared with placebo. And over 3 months both doses of the drug had significant benefit beyond placebo in terms of time-to-sleep onset.

If we look at total sleep time, again, the drug resulted in approved sleep duration that was maintained for 3 months of nightly use.

And wake after sleep onset had a very similar differentiation of the active drug from placebo.

As I mentioned, there's an immediate release and a modified release, and looking at a variety of studies that have been presented mainly in abstract form at this point, you can see that both formulations have improvements in terms of sleep latency, total sleep time, wakefulness after sleep onset, number of awakenings, sleep quality, etc.

As you might expect, comparing the immediate-release with the modified-release formulation, the sleep efficiency improvement is seen with the modified release; it makes perfect sense that you have more drug released later in the night so you're improving sleep efficiency as well.

In terms of daytime effects, that's also being investigated with indiplon, and in terms of overall insomnia severity, the drug seems to have a beneficial effect. It also improves self-reports of functioning during the daytime on measures such as alertness, quality of life, etc. So this medication is being assessed on a wider range of outcomes than has traditionally been associated with hypnotic medications, and it seems to have beneficial effects on both nighttime and daytime symptoms.

Another new drug that is under development is a controlled-release form of zolpidem and this, again, has a modified formulation such that there are immediate- and sustained-release portions in it. It has the same half-life as the parent compound because that's inherent to the drug itself, but some of it is released more slowly. And as you might expect, wakefulness after sleep onset is improved a bit further into the course of the sleep period compared with the regular formulation of zolpidem.

Let's turn to some of the other drugs used to treat insomnia, starting out with antidepressants. There is evidence that some of the sedating antidepressants do in fact have beneficial effects on sleep.

There is a poster looking at doxepin, a tricyclic antidepressant, in doses of 3 mg and 6 mg for insomnia, and it seemed to have some efficacy.

This is looking at a slightly higher dose — 25 mg or 50 mg, I'm not exactly sure. This is a study done in Germany and the basic notion here is that if you look at sleep efficiency measured polysomnographically, after 4 weeks of treatment, there was a benefit of doxepin over placebo. On discontinuation, those effects were lost as you would expect; it didn't look like there was a huge rebound. In terms of subjective sleep quality, again, there was an improvement in favor of the doxepin vs placebo.

I use doxepin in some of my patients, but the main antidepressant that is used is trazodone. Depending on how you count prescriptions, trazodone either is the first or the second most widely prescribed agent for treatment of insomnia in the United States. The issues are that it's not approved for that indication, and there have never been any dose-ranging studies to identify the optimal dose for this indication. We don't know how long it works and we don't really know too much about its safety in the setting of insomnia per se. We know about its safety when used in higher doses to treat depression.

These are data showing the longest study and the largest study of trazodone to date; this is a study that compared trazodone, zolpidem, and placebo for 2 weeks using self-report outcomes. In terms of subjective sleep latency, the trazodone and zolpidem groups improved and stayed improved for the second week; placebo started to catch up because the placebo-treated people were improving. With regard to subjective sleep duration, there was improvement in both active drugs that was maintained into the 2 weeks. The reason that you lose statistical significance at week 2 is not because the drugs wore off, but because the placebo group actually took off. That's the best evidence that we have regarding trazodone efficacy.

Antipsychotic drugs are being used as well, and I personally find this to be a bit of a disturbing trend because these medications, while undoubtedly very sedating, do have a range of side effects that can be potentially serious. There are 2 main drugs that are used this way.

Olanzapine is a drug that has a very complicated mechanism of action because it affects all sorts of different receptors. It's doing things to serotonin, cholinergic, and histaminic receptors, so we don't know exactly what is responsible for the sedative effect — likely some combination of all of these things. It has a long half-life, which is worth keeping in mind for a hypnotic drug, and there are very few studies that have looked at its effects on sleep per se. There is a study that was conducted in healthy subjects that showed increased slow-wave sleep, decreased wakefulness, so it had sedative-like properties. But keep in mind that this is a drug that can potentially have serious side effects, both acutely and over the long-term. One of the things that many patients on this drug are concerned about is weight gain and the possibility of developing portions of the metabolic syndrome. So you have to weigh in the balance here the fact that you have a drug with some pretty significant toxicities; it's not approved for this indication, and I'm not so keen on it.

The other one that is commonly used is quetiapine. Quetiapine has a shorter half-life than olanzapine and in that regard, it may be preferable. It has a shorter half-life; that's good. It has a very complicated mechanism of action. It has similar sorts of toxicities so, again, it is a little bit concerning.

There actually now is a sleep study that looked at the effects of quetiapine, and indeed it does seem to improve sleep efficiency, total sleep time, and subjective sleep quality. So it may well have those properties, but my caution to you is that it hasn't been assessed in people with insomnia per se and it does have some pretty substantial toxicities.

Anticonvulsant drugs are a big class; there are a lot of drugs that we could talk about, but I'm going to talk about tiagabine.

Tiagabine has an interesting mechanism of action. This is a GABAergic synapse, a subtype of a benzodiazepine GABA receptor. This shows the postsynaptic receptor that I showed you before.

Tiagabine is interesting because it is a GABA transport reuptake inhibitor, so it inhibits the GABA transporter 1 that sucks up the GABA from the synapse. By blocking that, it keeps more GABA in the synapse and more GABA can reach the postsynaptic receptors and have their effect. It's an interesting mechanism of action. It has a half-life that's in the range that we would like to see from a hypnotic drug, but you've probably seen recent reports that tiagabine has been associated in a small number of cases with new-onset seizures in people who didn't have seizure disorders. So you have to weigh that in the balance, too.

There are some empirical data on the effects of tiagabine on sleep in healthy older adults. Here we have placebo and tiagabine, and sleep efficiency is improved, stage 1 sleep is decreased (that seems like a good thing), and stage 3/4 sleep generally is increased by this drug; not significant, but in general an increase. Again, there are some properties that make it look like an acceptable hypnotic, but it has not been very well assessed yet.

Finally, we'll talk about some over-the-counter (OTC) and herbal preparations.

The most common OTC drugs used for treating sleep and one of the most common treatments overall are antihistamines. And among those, diphenhydramine and doxylamine are the most common ones. You would probably be as shocked as I was if you try to look up studies that support the efficacy of diphenhydramine for treatment of insomnia. The studies basically are just not there; you have to go back many years to find any studies at all and those studies in no way have the kinds of methodologies that we use to assess hypnotic drugs today in terms of assuring ourselves of their efficacy and safety. Keep in mind, too, that diphenhydramine not only is an antihistamine drug, but it's an anticholinergic drug and this can pose problems in our older patients.

Things like dry mouth and constipation may not sound like a big deal, but to many patients, they are a big deal. Urinary retention can constitute a medical emergency; delirium also is possible with anticholinergic drugs. A colleague of mine at the University of Pittsburgh, Pittsburgh, Pennsylvania, has done some studies in older adults in which he looks at cognitive function as related to serum anticholinergic load. So you can just measure anticholinergicity in a person's blood, and one of the main ways that it gets there is through diphenhydramine. And there's a clear relationship between how much anticholinergic drug you have in your bloodstream as an older adult and your cognitive functioning. It's not that more anticholinergic is good. So these drugs can have real-life impairments in terms of cognitive function.

I also was shocked because I looked up doxylamine, which I'm not as familiar with, and the only reports I could find in Medline were case reports of rhabdomyolysis and renal failure from people who had overdosed. The general assumption in the public is, "I can buy it at my grocery store; therefore, it must be efficacious and it must be safe." I think that we should just be careful with these drugs because they are not necessarily either of those things.

Valerian has been widely used and there are some data that relate to valerian use that I'll show. This is the ground-up root of a plant, and one of the problems with ground-up roots of plants is that they contain all sorts of chemicals. And the problem with valerian is that there are about 60 or 70 potentially active chemicals that may relate to its psychotropic effects and we don't really know what they are; they have names like sesquiterpenes and valepotriates. We don't know the active ingredients. The problem with all these things, too, is that when you send your patient to buy some, what brand do you recommend? They're unregulated and we don't know. There have been studies that have looked at the chemical composition of these and other agents and suffice it to say that you can buy valerian that contains no measurable valerian. That's kind of a "buyer beware" situation. Anyway, we don't really know how it works. Valerian root actually contains some GABA, but that's not how it works because GABA doesn't cross your blood-brain barrier. It does have sedative action, though, in animals and humans and there actually are some data.

I'm not going to go over this in detail, but valerian compared with placebo does seem to have some effects on self-report sleep. And there are polysomnographic (PSG) studies as well that show mainly a decrease in sleep latency; some studies show a little bit of increase in stage 3 and stage 4 sleep.

Another problem is that placebo is compared with valerian and it's also compared with valerian plus hop extract, so it's available in a combination preparation.

Melatonin is popular and is attractive to many patients because it is a hormone that your brain produces so it probably can't be too bad for you. And on balance, the adverse effects reported with melatonin have generally been pretty mild. There are persistent concerns that some people have vasoconstriction, reproductive effects, but no real clear signal emerging from the population on that.

In terms of efficacy, a recent meta-analysis was conducted looking at 17 trials of melatonin. The mean sleep latency reduction across these studies was 4 minutes, sleep efficiency increase was 2.2%, and sleep time increase was 13 minutes. So, pretty mild effects. There may be some cases in which this is useful, but by and large, in my patients, I haven't found melatonin to be really useful in most cases.

One final word about melatonin: keep in mind that it may have some mild hypnotic effects, but it also has some chronobiotic effects, which means that it is a phase-shifting agent. You can think of melatonin as "antilight," so it does to your circadian system what light does, only 12 hours apart. Melatonin given in the evening tends to advance circadian rhythms, and melatonin given in the early morning hours, if you ever happen to prescribe it that way, would delay circadian rhythms.

There are a few other things that we can talk about in terms of new and emerging treatments.

Here's the GABA transporter. There are other agents that work at extrasynaptic GABA receptors; this is the postsynaptic GABA receptor that we're used to talking about. But there are extrasynaptic GABA receptors as well that catch some of this GABA from the synaptic cleft.

One of those agents is gaboxadol. This is a structural analog of GABA, but it acts as kind of a nonspecific GABA-receptor agonist. It has a short half-life, is excreted by the kidneys, which is kind of interesting and different for a hypnotic-type medication, and it seems to be very different in terms of animal administration models; animals do not become cross tolerant between benzodiazepines and gaboxadol. They're doing different things; one does not substitute for the other. It's different than the new nonbenzodiazepine, but benzodiazepine-receptor-agonist drugs are acting very differently.

There are a variety of published results in animals and in humans; this is a study in healthy young adults, not insomniac patients. Gaboxadol compared with placebo improved sleep efficiency and also increased slow-wave sleep, and that does seem to be a fairly consistent finding with this medication, that it increases slow-wave sleep. Notice that it didn't have an effect on sleep latency, but these were healthy young adults, so it may have been pretty difficult to demonstrate any change anyway.

There's a set of drugs called alpha-2 delta ligands and you have been prescribing one of them without even maybe knowing that. Gabapentin (Neurontin) is one of these drugs. For a long time, people had no idea how gabapentin actually worked in the nervous system and now we're pretty certain that the way these drugs work is that they bind to voltage-sensitive calcium channels. There's a specific subunit that they bind to and these voltage-sensitive calcium channels, when they're activated, release excitatory neurotransmitters such as glutamate. So by inhibiting that channel, you decrease the release of excitatory neurotransmitters. It's a novel mechanism of action, something different. These drugs also have renal excretion, so that's different, and the half-life is in the range that we want. One of the main ways that pregabalin and gabapentin differ is that gabapentin has nonlinear pharmacokinetics, that is, at higher doses, less and less is absorbed so that when you increase the dose above about 1800 mg, it has a very small incremental effect.

There is a recently published study of pregabalin vs alprazolam in healthy adults, and pregabalin and alprazolam both decrease sleep latency, both improve sleep efficiency, and both improve sleep time. But what the pregabalin did that the alprazolam (Xanax) did not do is increase stage 3/4 sleep. So it has kind of a different signature than we would typically associate with hypnotic medications.

Ramelteon is another compound under development. This is a melatonin-receptor agonist and it's very selective for melatonin-1 and melatonin-2 subtype receptors. It actually has a higher affinity for melatonin receptors than even melatonin, and it has an active metabolite that also has affinity for melatonin receptors, so it's kind of melatonin on steroids. And it doesn't seem to do anything to other receptor types.

There are now published data looking at this compound in transient insomnia in adults, and with 16- and 64-mg doses, PSG latency to persistent sleep was decreased with this compound compared with placebo. Subjective sleep latency also tended to be improved at both of the higher doses. There have been some results showing that there were some effects as well on sleep continuity during the night, but the largest effect seems to be sleep latency.

The final new thing I will tell you about is 5-HT₂ (serotonin-2) receptor antagonists. These drugs have been looked at for a long time as possible anxiolytics, antidepressants, and sleep-promoting agents. And there are a lot of drugs that actually do antagonize 5-HT₂ receptors. Drugs such as trazodone, mirtazapine, doxepin, amitriptyline—all those sedating antidepressants have at least part of their mechanism explained by 5-HT2 antagonism. What's different about some of the newer agents is that they're relatively more specific for this mechanism. Unlike the tricyclics, they don't have a lot of antihistaminic and anticholinergic effects as well.

These are published data; this is a 5-HT₂ antagonist. To be honest, I don't know whether this is one that's under development as a hypnotic, but it does things that have a hypnotic-like profile, such as decrease stage 1 sleep. Actually it's a little bit different; it doesn't have the traditional effect on sleep latency, but it does do things such as increasing slow-wave sleep, it had some mild effects on rapid eye movement (REM) sleep as well. But sleep efficiency, total sleep time, and sleep latency, the usual hypnotic-type properties, were not affected by this drug.

How does all this fit into our clinical practice?

Going back to behavioral treatments, the real emphasis that we've seen in the past couple of years is trying to develop forms of cognitive behavioral therapies that are shorter and more focused, that use other modes of administration including group treatment setting, brief treatment durations of 2 or 4 sessions. There have been studies of Web-based, telephone-based interventions, all of which may help to disseminate these treatments to the populations that need them. Of course getting the word out, disseminating these treatments into the population is really the key, and in the future, we need to be creative about how we can get these efficacious treatments into the hands and homes of people who need them.

With regard to pharmacologic agents, I think that the prevalence data on insomnia has kind of hit home with many groups, including industry, and there's a large potential market here. The fact is we are seeing new drugs being developed, we're going to see more, and the exciting thing about that is that it gives us a lot more therapeutic options for our patients, which I think is a really good thing.

With regard to benzodiazepine-receptor agonists, we're seeing fine-tuned pharmacokinetic profiles, modified-release preparations that give us better coverage during the night, more flexibility. We've been seeing studies with longer durations of administration, which I think is useful — not that every patient should have long-term administration of a hypnotic drug. But for some people with chronic insomnia, it's appropriate and it's good to know how well or not those kinds of treatment work. New populations; primary insomnia is an interesting model, but it's not terribly realistic in terms of the population, so more and more studies are starting to look at treatment of insomnia and comorbid conditions, and I think that's a very positive development.

As I've told you, we have a lot of new molecular targets to look at, including the receptors that you probably learned about for the first time. There are all these different mechanisms of action not only hitting GABA and benzodiazepine receptors, but things such as the voltage-sensitive calcium channel, melatonin, 5-HT₂C, etc. Some of the new drugs have different profiles than current hypnotics, so they're not necessarily drugs that affect sleep latency. They're drugs that increase slow-wave sleep in some cases. They're drugs that may affect continuity in the middle of the night, but not necessarily sleep latency, or conversely, they may be drugs that affect only sleep latency. So there are a lot of different profiles and it's going to be incumbent upon us to get to know these drugs so that we can again apply them most effectively to our patient population.

The last thing I'll say is that there are still a lot of unanswered questions. We have these different sleep profiles that I was just mentioning, but we don't really know the consequences of that. Does it matter to our patients, clinically speaking, if we increase their slow-wave sleep? I don't know. It sounds like a good thing, but I don't know if it really matters. How does the efficacy and tolerability of these different agents compare? That'll be an interesting question. I think a very important question is, which patients will respond best to which agents? How can we best match patients to treatment? That's a great question and I think, unfortunately, it's going to be awhile before we have great answers to it. Finally, something that's actually pretty simple but we really don't have is an empirically based treatment algorithm for insomnia. We don't have a step-by-step algorithm that says first do this; if that fails, do this; and if that fails, do this or this. We need to develop that over the next few years.