This column extends discussions of several topics from previous issues, including case studies, 1 dose-response curves, 2 and the effect of the human genome project on drug development in psychiatry. 3 The idea for this column developed when I was recently asked to review an excellent article by Pierre and colleagues on the use of high-dose antipsychotic drug therapy. 4 The focus of that review article was whether there were adequate data to support the use of high doses of antipsychotic medications to treat the 30% or more of patients with schizophrenia and related psychotic disorders who are not adequately helped by the usual recommended doses outlined in the package inserts for the different antipsychotics. 5 To answer this question, Pierre and colleagues carefully reviewed the available data on the use of high doses for each of the available “atypical” antipsychotics. They concluded that modest but insufficient data were available to support the value of this approach.
While the review was thoughtfully done, the authors did not provide a theoretical discussion concerning whether there is a rational basis to guide clinicians when they consider such a strategy. In this column, I will therefore consider the following questions:
- How is the usually recommended dose range determined?
- Is there a rationale for exceeding the usually recommended dose?
- Does that rationale apply to all patients or only a select number?
- If the latter, is there any way to detect such patients?
How is the usually recommended dosage range determined?
The package insert for a drug contains the recommended dosage range. As with everything in the package insert, that range is jointly determined by the pharmaceutical manufacturer and the Food and Drug Administration. It is based on the human clinical trial experience with the drug and represents the dosage range associated with maximum benefit and minimum risk of adverse effects in the usual patient in the registration clinical trials of the drug. 6
To put this matter into perspective and relate it to the issue of whether doses higher than the usual range might be appropriate for nonresponders, the following two equations should be helpful:
Even the occasional reader of this column is likely to be familiar with these equations. Equation 1 outlines the three variables that determine response to any drug: the first variable determines the potential actions of the drug on the body; the second determines whether sufficient drug reaches the site of action to engage the desired mechanism of action to a sufficient extent to have the desired effect; and the third is the biological variance that can shift the dose-response (or more precisely concentration-response curve) in a specific patient relative to the “usual” patient.
The second variable in Equation 1 is determined by the dosing rate relative to the clearance in a specific patient as represented in Equation 2. This second equation illustrates that the clearance in a specific individual is as important as dose in determining that individual’s drug response.
Clinical trials are essentially population pharmacokinetic studies in which the goal is to determine the usual dose needed relative to the “usual” human clearance to achieve a concentration that will engage the right mechanism of action (variable 1 in Equation 1) to the right degree (variable 2 in Equation 1) to produce the desired clinical effect. In other words, variable 1 multiplied by variable 2 determines the mass action of the drug in the usual patient enrolled in the registration trials.
Where does the determination of the usual dosage range begin?
As explained in the series of columns on drug development in psychiatry and the human genome project, 3 determination of the usual dosage range begins with drug discovery and is further refined throughout a drug’s development, but particularly in phase I studies. In drug discovery, the goal is to develop a new drug with either a novel or a refined mechanism of action (variable 1 in Equation 1). Today, this task is frequently done by selecting a target based on our improved (but still with a long way to go) understanding of the pathoetiology or pathophysiology of the disease process of interest.
If Alzheimer’s disease, for example, is caused by the excessive accumulation of beta amyloid, then a drug that inhibits the production of this protein could prevent the disease. The isolation of the enzyme mediating the production of this metabolite would thus provide a novel target for drug development. One approach would be to screen a variety of compounds against the intrinsic activity of such an enzyme to determine the structure-activity relationship needed to inhibit the enzyme and thus prevent the production of beta amyloid. Another would be to actually crystallize the enzyme and directly determine the structure and conformation of the active site and deduce the necessary structure-activity relationship based on that knowledge.
Once that is accomplished, the next step would be to synthesize a series of molecules having that structure-activity relationship and then test them for their potential suitability as drugs (e.g., dissolution characteristics) and also screen them against other known but undesired targets. The next step is to further tweak the structure so that the molecule is selectively active at the desired site but not at undesired sites. The goal of this process is to reduce the potential for the new molecule to cause any undesired effects at any clinically relevant concentration, while, at the same time, ensuring that the new molecule has as great a binding affinity for the desired target as possible. The advantage here is that even low concentrations of the molecule will be effective.
As illustrated in Equation 1, the binding affinity of the drug for the target determines the concentration that must be achieved at the target to engage the mechanism to a physiologically relevant degree. The plasma concentration of the drug that must be reached to achieve the necessary concentration at the site of action is determined by any relevant partition coefficients between the central compartment (circulating blood) and the compartments of interest (presumably the brain, in the case of CNS drugs).
The isolation of the genes that code for drug metabolizing enzymes and other clinically relevant pharmacokinetic mechanisms (e.g., drug transporter proteins) now permits a much better estimate of the dose of the drug that will be needed in man to achieve that plasma drug concentration and, more importantly, the concentration at the desired target.
Preclinical animal studies further test whether the drug does what it was intended to do and does not do anything that was unintended and potentially adverse. More sophisticated animal models based on a better understanding of the pathoetiology and pathophysiology of the illness of interest permit better determination of whether the original hypotheses that led to the discovery of the potential new drug are likely to be valid.
The goal of phase I studies is then to extend this line of research to human subjects and specifically to determine the safety, tolerability, and pharmacokinetics of the new molecular entity. The question is: “Can the entity be a drug in humans”? In other words, can it reach the site of action in sufficient concentrations to be effective to treat the target illness? The answer to this question is dependent on both the pharmacokinetics of the drug in humans as well as its safety and tolerability.
The first question is whether it is physiochemically possible for the drug to be absorbed into the systemic circulation and reach (i.e., be distributed to) the desired target to a physiologically relevant degree. The second is whether the drug at that concentration is sufficiently safe and well tolerated so that it is clinically possible to reach such concentrations. For example, if the drug causes a serious toxicity problem in the heart (e.g., arrhythmia) at a lower concentration than is effective in the brain, then it is irrelevant whether it is theoretically possible to achieve such concentrations in the brain. The same is true for less serious but poorly tolerated adverse effects such as nausea and vomiting.
One goal of phase I studies, therefore, is to determine the optimal dose for subsequent phase II and III efficacy trials; this goal can be achieved in several ways. An ideal method is to have a surrogate marker for the desired concentration, for example, the use of brain imaging such as positron emission tomography (PET), to determine whether a predetermined occupancy rate of a specific receptor target in the brain can be achieved. 7,8 Another approach is more empirical and involves using escalating doses to determine the maximum safe and well-tolerated dose. Plasma drug concentrations are also measured and related back to the predicted concentration needed for efficacy, as explained above.
Phase I studies are usually done with young, healthy, normal volunteers. However, patients with psychiatric illness may be more—or less—sensitive to the effect of a drug; therefore, “bridging” studies are now increasingly being done in individuals who more closely approximate the eventual population that will be treated with the drug. An example of such a bridging study would be testing a potential new drug for dementia of the Alzheimer’s type (AD) in healthy, elderly volunteers, since the eventual target population for drugs with this indication will be patients over the age of 65. Older patients are frequently more sensitive to the effects of drugs—hence the clinical adage to “start low and go slow” in elderly patients. That may be because elderly patients are intrinsically more sensitive to the mechanism of action of the drug (variable 1 in Equation 1) or because they generally have lower clearance and thus need lower doses (Equation 2).
Studies might also be done in elderly patients with mild cognitive impairment. The advantage of this approach is that it gets still closer to the eventual target population and it also has the potential to test, early in the development of the drug, whether it is likely to be efficacious. 9
Bridging studies can also be done with volunteers who are mildly symptomatic with a variety of target psychiatric illnesses, since the illness itself (variable 3 in Equation 1) may make these individuals more or less sensitive to the desired or undesired effects of the drug. For example, there is some evidence, principally based on experience in drug development rather than published findings in the literature, that patients with schizophrenia can tolerate higher doses of dopamine-2 (D2) receptor blockers than normal controls without developing acute extrapyramidal side effects (EPS). That may be either a basic feature of the pathophysiology of the illness or possibly the result of developing tolerance from chronic exposure to D2 blockers.
Such bridging studies involve volunteers who are mildly and/or chronically symptomatic with affective and anxiety disorders as well as schizophrenia. In the latter case, the volunteers usually have mainly deficit symptoms but may also have low grade and stable positive symptoms. 10 These patients may be on no medications, only on anti-anxiety medications, or on low doses of antipsychotic medications (often below the lower limit of the usual recommended dose range). They are admitted to a secure research unit for the duration of the study. All are able to and do give informed consent to participate.
Once the maximum tolerated dose is established, it is rarely, if ever, exceeded in subsequent clinical trials with the drug (e.g., the phase II and III studies that are done for registration purposes). Instead, phase II and III studies test the drug within the range defined in the previous stages of the drug development process. If its benefits at such concentrations can be adequately demonstrated and outweigh any adverse effects relative to the seriousness of the disease being treated, then the drug is approved and the package insert reflects the dosing range that was tested in the phase II and III studies.
The package insert rarely, if ever, mentions the maximum tolerated dose or any of the adverse results experienced on higher doses in the usual phase I or “bridging” studies. These studies are also rarely published since the results are primarily useful only in setting the stage for the phase II and III studies and thus are generally perceived as not being of interest to prescribers.
What happens when the drug is marketed?
Ironically, one of the most common questions posed by clinicians when a drug first enters the market is whether they can prescribe higher doses than those recommended in the package insert. They ask this question for the same reason that led Pierre and colleagues to review the literature on the efficacy of higher than recommended doses of antipsychotic medications in patients who have not benefited from a trial at the upper end of the usually recommended dose range. Any clinician who treats even a modest number of patients with schizophrenia and related psychotic disorders will encounter a significant number of such patients. In this situation, the clinician has several options:
- switching to another antipsychotic medication,
- adding an augmenting agent (frequently another antipsychotic medication),
- using higher than usually recommended doses
- settling for less than an optimum outcome.
Option 3 is a natural strategy for many prescribers because they are trained to adjust dose on the basis of clinical assessment of response. 11 As explained in an earlier column, 12 such dose adjustment is almost always upward when the clinician is faced with inadequate response and no adverse effects, since those two observations are usually taken to mean that an adequate concentration of the drug is not reaching the desired site of action (variables 1 and 2 in Equation 1).
However, at least two caveats need to be made. First, the drug may not be working on the right site of action for that specific patient with his or her specific pathophysiology. In this case, there is no concentration that will produce the desired benefit. This caveat results from the fact that psychiatric illnesses are syndromic diagnoses and thus are likely to be more than one illness when understood from a pathophysiological or pathoetiological perspective. 13 Second, as discussed in several earlier columns, 11,14–17 the adverse effects of selective CNS drugs can ironically present as inadequate response. In this instance, increasing the dose is precisely the wrong course of action and can lead to more undesirable outcomes than simply a lack of response. 11
Readers should keep these two caveats in mind as they read the rest of this discussion, which will focus on why the drug concentration may be inadequate and why upward dose adjustment might be helpful in specific patients. In every instance where dose escalation is beneficial, it will reinforce the use of higher doses by clinicians the next time they encounter such a patient. Such clinical reinforcement is precisely the scenario that has led clinicians to publish the case series and small studies of various designs reviewed by Pierre and colleagues. 4
Clinical trials are population pharmacokinetic studies
The truth of the statement that “clinical trials are population pharmacokinetic studies” is obvious from the discussion of how the usual dosing range is determined. In the clinical trial, the investigators are trying to determine the usual dose for the usual patient enrolled in the trial that will affect the primary target(s) to a sufficient degree to produce the desired clinical outcome (e.g., relief of positive and negative symptoms of schizophrenia).
The problem is that the usual patient in the clinical trial may not be the usual patient in the clinician’s practice. 18 For this reason, the dosage range identified in the clinical trial may not be appropriate for every patient in clinical practice, such as the patient with treatment-refractory schizophrenia. Such patients are virtually always excluded from registration trials because of the concern that they will likewise not respond to the new drug and thus will reduce the power of the study to detect a drug-placebo difference.
Based on Equation 1, there are several reasons why a patient might not respond to the usual dose but would respond to a higher dose. The simplest explanation is that that patient clears the drug faster than the usual patient and hence does not achieve the usual concentration needed to engage the drug’s site of action to a sufficient degree to produce a clinically meaningful response. Such faster clearance could be genetically determined or might result from co-administration of a second drug that is capable of inducing the metabolism of the first medication. 19
There are other genetic explanations. These include mutations in the regulatory protein(s) that are the site(s) of action for the drug, so that a higher concentration is needed to engage those site(s) of action to a physiologically relevant degree, or mutations in regulatory proteins, such as ABC transporter proteins, that determine the concentration of the drug that reaches the site of action. For example, mutations in the serotonin transporter protein have been reported that can alter the response rate to serotonin reuptake pump inhibitors. 20–22 A number of other mutations in CNS regulatory proteins that have the potential to be clinically relevant have also recently been reported. 23–25
Is there any way to determine which is the correct explanation in any given case?
Therapeutic drug monitoring (TDM) can be used to determine whether a patient has faster than usual clearance of a drug, using Equation 2. If one knows the dose the patient is taking and the concentration achieved on that dose (from TDM), then one can determine the patient’s clearance by re-arranging Equation 2 as follows:
If the level is lower than was usually achieved by patients in the clinical trials, then that is the simplest explanation for why the patient is not responding. In routine clinical practice, however, there are two possible explanations for lower than usual drug levels. One is faster than usual clearance; the other is that the actual dosing rate is not the prescribed dosing rate, because the patient is not being compliant with the prescribed dosing rate.
If compliance is the problem, then trying a higher dose may be futile or could lead to untoward results if the patient decides to become compliant or if the patient has been hospitalized because of a relapse and thus becomes compliant because the medication is being given by the nursing staff.
TDM can also be used to determine whether the patient has been noncompliant by obtaining a pharmacokinetic profile after a single dose. However, this approach is usually cumbersome and considered expensive in routine clinical practice. A simpler way to check for compliance is to obtain a repeat level after ensuring compliance by having the dosing supervised by a trustworthy individual such as a concerned family member, visiting nurse, or case manager. The dosing duration should be five times the half-life of the drug at a stable dosing rate. If the repeat level is comparable to the previous level, then rapid clearance is essentially confirmed. In this instance, the use of a higher than usual dose is appropriate because dose is only one of two variables that determine drug concentration and, in this case, the dose must be adjusted to compensate for the unusually rapid clearance.
While TDM is useful in answering questions about clearance and compliance, it cannot determine whether the problem is a mutated receptor resulting in altered binding affinity, or a mutation in the mechanism controlling distribution of the drug to the target compartment that results in an inadequate concentration being achieved at the target despite adequate plasma concentrations. Such altered distribution could also be the result of a drug-drug interaction. Altered drug distribution into the peripheral compartment of interest (e.g., the brain) could be determined by PET scanning or by measuring the concentration of the drug in cerebrospinal fluid. However, these approaches are used only in a few research sites and are not currently feasible for use in routine clinical practice.
Without being able to determine whether the problem is altered binding affinity or a mutation in the distribution mechanism, the clinician will want to cautiously weigh the potential risks and benefits of trying higher than usual doses if TDM indicates that the patient has usual plasma concentrations and hence usual clearance. In this situation, the clinician will want to consider the therapeutic index of the drug and the extent of the clinical database supporting the safety and tolerability of higher doses. With respect to antipsychotic medications, the review by Pierre et al. 4 is not encouraging. They found the extent of the database to be quite modest and the reported results not encouraging. In addition, the design of the studies was often less than ideal. For example, some studies lacked a control group for time on drug and for nonspecific (i.e., “placebo”) response. The absence of a control group might seem irrelevant when dealing with patients with treatment-refractory illness but the interested reader is referred to an earlier column that discussed open-label trials of an alternative serotonin reuptake inhibitor (SRI) in patients who had historically not responded to a previous SRI. 26 It is important to recall that placebo treatment is not nothing but rather all the therapeutic components except the intervention that is being formally studied. Without a rigorous design, it is easy to erroneously attribute benefit to the agent being studied and to discount the contribution of all the rest of the care being provided. When the issue is exceeding the recommended dosing range, the consequences are likely to be too important to rely on designs that are prone to such misinterpretations. A subsequent column will discuss why the database on use of higher than usual doses of antipsychotics is inadequate and what should be considered when doing or evaluating such a trial.
The purpose of this column was to provide a conceptual framework for thinking about the possible appropriateness of using higher than usual doses in carefully selected patients and to explain how TDM could be used to identify such patients. However, this column is not intended to advocate such use. Instead, it is a call for more research using these concepts and TDM to empirically test whether higher than usual doses can be safely and effectively used in patients who are refractory because of unusually rapid clearance.
1. Preskorn SH. The case studies section. Series of columns published in the Journal of Psychiatric Practice (www.preskorn.com/column1.html
2. Preskorn SH. The rationale for and clinical implications of dose-response curves. Series of columns published in the Journal of Psychiatric Practice (www.preskorn.com/column3.html
3. Preskorn SH. Modern drug development and the Human Genome Project. Series of columns published in the Journal of Psychiatric Practice (www.preskorn.com/column4.html
4. Pierre JM, Wirshing DA, Wirshing WC. High-dose antipsychotic therapy: Desperation or data-driven? Current Psychiatry 2004;3(8) in press.
5. Conley RR, Buchanan RW. Evaluation of treatment-resistant schizophrenia. Schizophr Bull 1997;23:663–74.
6. Kinon BJ, Ahl J, Stauffer VL, et al. Dose response and atypical antipsychotics in schizophrenia. CNS Drugs. 2004;18:597–616.
7. Mamo D, Kapur S, Shammi CM, et al. A PET study of dopamine D2 and serotonin 5-HT2 receptor occupancy in patients with schizophrenia treated with therapeutic doses of ziprasidone. Am J Psychiatry. 2004;161:818–25.
8. Meyer JH, Wilson AA, Sagrati S, et al. Serotonin transporter occupancy of five selective serotonin reuptake inhibitors at different doses: An [11C]DASB positron emission tomography study. Am J Psychiatry 2004;161:826–35.
9. Wesnes KA, Preskorn SH, Friesen S, et al. NS 2330 enhances cognitive function in normal volunteers and volunteers with possible Alzheimer's disease (AD). Clin Pharmacol Ther 2002;71:17.
10. Reyes LF, Preskorn S, Khan A, et al. Pharmacokinetics (PK) of donepezil and risperidone in males with schizophrenia. Clin Pharmacol Ther 2001;69:44.
11. Preskorn SH. Polypharmacy in a patient with refractory major depression: Part I: The case. Journal of Psychiatric Practice 2002;8:370–6.
12. Preskorn SH. Relating clinical trials to psychiatric practice: Part I: The case of a 13-year-old on aripiprazole and fluoxetine. Journal of Psychiatric Practice 2003;9:307–13.
13. Preskorn SH, Baker B. The overlap of DSM-IV syndromes: Potential implications for the practice of polypsychopharmacology, psychiatric drug development, and the Human Genome Project. Journal of Psychiatric Practice 2002;8:170–7.
14. Preskorn SH. Do you believe in magic? Journal of Practical Psychiatry and Behavioral Health 1997;3:99–103.
15. Preskorn SH. The slippery slide. Journal of Practical Psychiatry and Behavioral Health 1999;5:50–5.
16. Preskorn SH. A message from Titanic. Journal of Practical Psychiatry and Behavioral Health 1998;4:236–41.
17. Preskorn SH. Multiple medications, multiple considerations. Journal of Psychiatric Practice 2001;7:48–52.
18. Preskorn SH. Relating clinical trials to psychiatric practice: Part II: The gap between the usual patient in registration trials and in practice. Journal of Psychiatric Practice 2003;9:455–61.
19. Preskorn SH. A tale of two patients. Journal of Practical Psychiatry and Behavioral Health 1999;5:160–4.
20. Arias B, Catalan R, Gasto C, et al. 5-HTTLPR polymorphism of the serotonin transporter gene predicts non-remission in major depression patients treated with citalopram in a 12-weeks follow up study. J Clin Psychopharmacol 2003;23:563–7.
21. Perlis RH, Mischoulon D, Smoller JW, et al. Serotonin transporter polymorphisms and adverse effects with fluoxetine treatment. Biol Psychiatry. 2003;54:879–83.
22. Lotrich FE, Pollock BG, Ferrell RE. Polymorphism of the serotonin transporter: implications for the use of selective serotonin reuptake inhibitors. Am J Pharmacogenomics 2001; 1:153–64.
23. Malhotra AK, Murphy GM Jr, Kennedy JL. Pharmacogenetics of psychotropic drug response. Am J Psychiatry 2004;161: 780–96.
24. Basile VS, Masellis M, Potkin SG, et al. Pharmacogenomics in schizophrenia: the quest for individualized therapy. Hum Mol Genet. 2002;11:2517–30.
25. Pickar D. Pharmacogenomics of psychiatric drug treatment. Psychiatr Clin North Am. 2003;26:303–21, vii.
26. Preskorn SH. The appearance of knowledge. Journal of Practical Psychiatry and Behavioral Health 1997;3:233–8.