While 30% is a disappointing response rate for any medication, this conversation made me think about more global issues in psychopharmacology. We have all had the experience of prescribing medication for a set of symptoms and getting completely different responses in different patients. Sometimes the medications work, and sometimes they do not, for the same complaints. Why is this?

Probably the greatest cause for the differential response to the same psychopharmacology is the lack of precision of psychiatric diagnosis. Our Diagnostic and Statistical Manual of Mental Disorders (DSM) is currently in a slight revision from its fourth version, and there is no question that each new version and each revision of each version has represented an improvement. It is important, however, to appreciate the DSM for what it is and what it is not. In its most fundamental use, the DSM allows us to talk to each other in such a way that we can be reasonably sure that we are all talking about the same thing. As such, when we talk about a specific disorder, such as paranoid schizophrenia or recurrent major depressive disorder, we already have a mutually agreed upon document that spells out relatively clearly exactly what those things are. One of the quantitative ways that we can measure improvement in the DSM is increasing amounts of interrater reliability. As the DSM improves, if two of us see the same patient in the same period of time, we are increasingly likely to make the same diagnosis.

The DSM is relatively scrupulous about avoiding statements about either etiology or treatment. As such, it is a completely phenomenologic document. It simply says what these disorders look like. It makes virtually no statements about pathophysiology, underlying psychology, or any kind of treatment recommendation. There is neither the data nor the mandate to do this in any comprehensive way. It is considerably easier to agree on diagnosis than to agree on etiology or treatment at the present time. While there are documents that do address these issues, such as textbooks and treatment guidelines, they are considerably more equivocal and often controversial. It is considerably easier to say, for example, what depression looks like than it is to say what causes it and what is the definitive treatment for it.

Choosing the right treatment is dependent at least in part on making the right diagnosis. While we have been able, for the most part, to agree on what psychiatric disorders look like, when different disorders look remarkably similar, diagnostic process becomes more unreliable and, as such, so does the efficacy of the selected treatment. We do not have reasonable biological markers for psychiatric disorders yet. We are working on them, and I hope to see at least some in my lifetime, but until the clinician in practice can have easy access to a quick and reliable definitive study that makes a psychiatric diagnosis, we are stuck with the lists of symptoms and signs in the DSM. If we miss something, or come to the wrong conclusion from the history and examination that we obtain, we run the risk of at least prescribing an ineffective treatment, and at worst prescribing a treatment that would exacerbate the patient's illness. Examples of this are prescribing stimulants for a child suspected of having attention-deficit/hyperactivity disorder but who actually has bipolar disorder or prescribing antidepressants without a mood stabilizer for a patient thought to have unipolar major depression but whose diathesis is really from bipolar disease and who begins to cycle more rapidly and become irritable on antidepressants alone.

Probably the most common reason for differential treatment response is undiscovered, or at least inadequately characterized, subtypes of the disorder. Clinical psychiatric research has been remarkably unsuccessful in describing subtypes of disorders that have relevance for treatment selection. There is some evidence that rapid cycling bipolar disorder responds better to anticonvulsants than to lithium and that atypical depression may respond preferentially to monoamine oxidase inhibitors, but even these subtypes do not have unequivocal data backing them up. Intuitively, it sounds almost ridiculous to say that all of the presentations that we see of depression are the same illness, yet that is for the most part how we describe them. Similarly, it does not sound sensible to say that all forms of depression would respond to the same medication, yet we continue to struggle to find the differences in the way that our patients present that would help us make decisions about which treatment to choose. For better or for worse, most decision-making for selection of a pharmacotherapy is based on side-effect burden. This is not to minimize the importance of this kind of decision-making, because no treatment will be effective if the patient is particularly sensitive to its side effects and cannot tolerate it. The problem is that even if they tolerate the treatment well, there is no guarantee that it will be effective. We have all seen patients who dismally fail their first medication trial and do beautifully on their second. We then think, "they must have had the type of depression for that particular drug." To date, we have not had much success in prospectively subtyping depression, or any other psychiatric illness for that matter, based on treatment response.

The lack of a precise diagnosis that is predictive of treatment response, while important, is not the only reason for treatment failures. Pharmacokinetics can play a large role in the success or failure of any given medication regimen. One of the concepts that we take for granted in all pharmacology is standardized dosing. If we choose a pharmacologic treatment for a certain disorder, we look up what the dose should be and prescribe that dose. On closer examination, however, the idea that there is one dose that is tolerable and effective for most or all of our patients may be considerably oversimplistic. People vary enormously in size and weight. Their bodies also vary enormously in fat and water composition, and these are where drugs tend to accumulate and distribute. Even if these factors could be addressed in the choosing of a dose, data continue to be collected about the enormous individual variation in patients' abilities to metabolize and excrete medications. Some generalizations can be made about some ethnic groups and whether they tend to be rapid or slow metabolizers, but even the extent to which ethnically similar people can vary in the rate in which they metabolize compounds seems to be much greater than we ever gave credit for and seems to imply that the task of choosing a dose for any given medication for any patient is quite daunting. Most experienced psychopharmacologists will have stories to tell about the patient who did extremely well on either a much smaller or much larger dose of medication than usual. Often, this will be discovered as the result of a medication error or some other serendipitous event. It is very difficult to discover these situations empirically.

One confounding variable for whether a medication is going to be effective has gotten a lot of attention, and that is the presence of other medications or agents, prescribed or otherwise, that may interfere with its effect. Much has been written about the importance of knowing and checking for drug interactions and, as such, it is not necessary to go into that in great detail here. However, a few points are worth making. Much has been written about the cytochrome P450 isoenzyme system and knowing which drugs induce or inhibit which isoenzymes. It is important to remember, however, that there are other drug interactions besides cytochrome P450 driven ones. For example, lamotrigine and valproate compete for glucuronidation and, as such, inhibit each other's metabolism, but this has nothing to do with the cytochrome P450 system. Another issue of drug interactions has gotten a lot of attention in the press lately but cannot be overstated. One must inquire carefully about over-the-counter food supplements or herbal compounds purchased at health food stores, as these can often undo whatever benefit prescribed psychotropics may have. Many clinicians treating bipolar patients, for example, have done more good by suggesting stopping Ephedra-based supplements that are increasing irritability and cycling than by prescribing another agent.

Finally, of course, there is the question of pharmacodynamics. Since we fundamentally do not understand how these drugs work, it is difficult for us to understand in any comprehensive way why they do not work. We do have some sense that whatever it is that they do therapeutically, they do it considerably far upstream from their targets. If, for example, a patient takes a serotonin reuptake blocker, their serotonin transporters are blocked within an hour. It takes considerably longer for their depression to resolve, even in the best of treatment responses. Many things must happen physiologically between when they start their medication and when they actually have an improvement. Since we are not sure what happens when it goes right, we are equally uncertain about where it might go wrong.