Email this   Print this

Relating Clinical Trials To Psychiatric Practice: Part II: The Gap Between the Usual Patient in Registration Trials and in Practice

SHELDON H. PRESKORN, MD

Journal of Psychiatric Practice, November 2003, 455-461

This column is the second in a two-part series. The first column presented the case of a 13-year-old boy treated with 60 mg/day of both aripiprazole and fluoxetine and discussed how the treatment in that case varied from the available clinical trial data.¹ It also discussed the multiple mechanisms underlying the drug-drug interaction that occurred and how they related to the clinical presentation.

This second column presents a more general discussion of the disparity between patients eligible for clinical trials and those seen in daily practice. Recognition of that disparity can help clinicians more readily interpret clinical trial data and use that knowledge in their practices.

The Equations

The following two equations are fundamental to understanding the safe and effective use of any drug:

  Effect = affinity for and intrinsic activity
              at site of action
         x concentration at site of action
             absorption, distribution, metabolism,
             elimination (ADME)
         x biological variance among patients
             genetics, age, disease, environment (GADE)
                                            (Equation 1)

and

  Drug concentration = closing rate / clearance
                                            (Equation 2)

Equation 1 covers the variables that determine the effect of a drug--whether desired or undesired, planned or unintended. To produce an effect, the drug must bind to and affect a regulatory protein (variable 1 in equation 1) in a specific way (e.g., agonism) to produce an effect. Variable 1 gives the drug the potential to have an effect but alone is not sufficient; instead, the drug must reach the site of action and reach it to a sufficient degree to engage it to a clinically relevant degree (variable 2 in equation 1). Variable 3 in equation 1 takes into account the modifying influences of interindividual biological variability (whether inherited or acquired) that can shift the dose-response curve in a specific patient relative to the usual patient. The key term in the preceding sentence is "usual," and that is a function of the population being sampled--which is the focus of this column.

Equation 2 illustrates the two variables that determine the drug concentration achieved in a specific patient: dosing rate (i.e., how much drug is going into the patient over a given unit of time) and clearance (i.e., how much drug the patient can clear from his or her body per the same unit of time). Parenthetically, 1 day is the most common clinically relevant unit of time, since most drugs are dosed once a day, but the relevant unit could instead be measured in minutes or hours or weeks, depending on the nature of the drug and how it is delivered. For example, the relevant unit is minutes for an intravenous drug being given to sedate a patient; whereas the relevant unit for a depot medication is weeks.

Drug Development and Clinical Trials

This column is also part of a series on the drug development process.² As explained in greater detail in other columns in that series, drug development is based on equation 1 as follows. The goal of the drug discovery phase is to select a target of interest relevant to a disease (variable 1 in equation 1) and then design a molecule to affect that target in a specific way: agonism, antagonism, inverse agonism, or somewhere between full agonism and full antagonism, or between full antagonism and full inverse agonism. Once the molecule has been synthesized, it enters preclinical drug development.³ In this phase, it is put through a variety of in vitro tests (e.g., receptor binding assays, humanized single cellular organisms) and in vivo tests (e.g., whole animals) to more fully determine its pharmacological characteristics. It then enters Phase 1 through 3 clinical trials.

Phase 1 studies are initially performed in normal young volunteers and may then be done in normal elderly volunteers or mildly symptomatic volunteers, depending on the intended clinical use of the drug. The goals of Phase 1 studies are to determine the safety, tolerability, and pharmacokinetics of the drug in these populations These outcome measures are always related to the dose of the drug and its route of administration (e.g., oral, acute intravenous, depot intramuscular). These routes of administration are related to the absorption characteristics of the drug, which is the first ADME characteristic listed under variable 2 in equation 1. Thus, the first clinical phase of drug development is designed to answer the question: Can the drug be given in such a way that a high enough concentration can be achieved (variable 2 in equation 1) to engage variable 1 to a clinically useful degree so that the efficacy of the drug for the condition being treated in subsequent Phase 2 and 3 studies is likely to outweigh any safety or tolerability problems caused by the drug?⁴

Phase 2 is made up principally of small, tightly controlled proof-of-principle (POP) studies and pivotal efficacy trials with the goal of providing sufficient data on efficacy to support a new drug application (NDA) to a drug regulatory agency such as the Food and Drug Administration (FDA) in the United States.

Phase 3 involves larger scale studies that more widely sample the range of humans who may receive the drug. The goal of this phase is to more closely approximate clinical practice to see if the drug’s efficacy is sufficiently robust to be seen in a larger number of multicenter studies that involve more investigative teams and perhaps less experienced investigators. Another goal of Phase 3 studies is to determine the safety and tolerability of the drug in both short studies of a few weeks and longer open or double- blind maintenance studies of 6-12 months or longer (variable 3 in equation 1).

In essence, clinical trials are population pharmacokinetic- pharmacodynamic studies.^5,6 The goal is to determine the usual dose(s) needed relative to the usual clearance characteristics of the drug and the usual pharmacodynamic characteristics in the population being studied. The usual clearance refers to the summation of the usual absorption, distribution, metabolism, and elimination characteristics of the drug in the usual human enrolled in these studies. Of course, humans are not all the same and these characteristics may differ substantially among some subpopulations, such as individuals who are genetically deficient in cytochrome P450 CYP 2D6 activity,⁷ which is but one of many genetically determined variabilities in pharmacokinetics among humans. Changes in the pharmacokinetics of a given drug can also be related to age, disease, and internal environmental. An example of a change related to the patient’s internal environment is the production of a phenotype of genetic deficiency of CYP 2D6 by the coadministration of the lowest usually effective dose of a drug such as fluoxetine or paroxetine.⁷

There may also be interindividual variability in the pharmacodynamics of a drug among humans. Such differences in dose-response can be due to any of the four variables listed under variable 3 in equation 1. For example, there may be genetic mutations of the target site (e.g., a receptor) that alter the binding affinity and/or the intrinsic activity of the drug at its site of action. Examples include mutations in beta adrenergic receptors that affect the efficacy of beta adrenergic blockers as antihypertensives and genetic mutations of the serotonin transporter protein that have an impact on the efficacy of the serotonin reuptake pump inhibitors as antidepressants and/or anxiolytics.^8-11 There can also be age-related or diseaserelated changes, such as the loss of dopamine neurons that occurs as a function of age or, more dramatically, as a function of Parkinson’s disease. The loss of these neurons can dramatically shift the dose-response curve involved in the production of acute extrapyramidal symptoms by dopamine-2 receptor antagonists such as conventional antipsychotics (e.g., haloperidol) and even "atypical" antipsychotics. The presence of co-prescribed drugs can also shift the dose-response curve either by affecting the same target or a related target. For example, the likelihood of an individual experiencing serotonin syndrome as a result of taking a serotonin uptake pump inhibitor would be increased if he or she was also taking a monoamine oxidase inhibitor or, conceivably, had a genetic mutation that resulted in either low or no activity of this enzyme, and such individuals do exist.^12-15 As mentioned above, drugs are an acquired form of biological variance that can mimic a genetic form of such variability.

The Goals of Registration Clinical Trials

At least two, and possibly three, parties are involved in the clinical trials done to register a drug. One is obviously the pharmaceutical company that is developing the drug. Another is the regulatory agency that has the responsibility for approving the drug for marketing. That regulatory agency represents the consumer and the physicians who will respectively take and prescribe the drug. The involvement of physicians may be relatively informal (e.g., experts serving on advisory panels) or may be more organized and proactive, as has been the case in some instances in U.S. psychiatry, such as when organized psychiatry lobbied for the testing and approval of lithium and clozapine in the early 1970s and 1990s, respectively. In general, organized psychiatry has taken a less active role in the drug development process than have oncologists and infectious disease specialists, especially those treating human immunodeficiency virus. Thus, psychiatric drug development in the United States has mainly reflected the concerns of manufacturers and the FDA.

The goal of the pharmaceutical company is to register the drug for marketing and to do so in as financially responsible a manner as possible. The goal of the FDA is to determine whether the benefits outweigh the risk relative to the condition being treated. The goal for the clinician is to know where to place the drug in the overall context of the available options and to know for which patients and situations the drug is most efficacious and safe and in which patients and situations the drug should be avoided. The problem is that current registration trials often do not fully meet the goal of the clinician for a number of reasons.

The Dilemma for the Company

As discussed in previous columns, psychiatric conditions are currently understood at the level of syndromes rather than at the level of pathophysiology or pathoetiology.¹⁶ Disorders such as major depression, schizophrenia, and bipolar disorders may well be more than one condition when understood from the more sophisticated perspectives of pathophysiology and/or pathoetiology. This may well account for the large noise-to-signal ratio seen in clinical registration trials of drugs such as antidepressants, in which it is common to see placebo response in one third of the subjects, drug-specific response in one third of the subjects, and nonresponse in one third.¹⁷

The downside to this situation is that the studies have to be large enough to have sufficient power to overcome the poor noise-to-signal ratio. The upside is that the potential market for the drug is everyone who meets the clinical criteria for the syndrome because there is currently no way to know who will respond and who will not--other than giving them an empirical trial of the drug. Such a trial can take weeks, during which the drug is often escalated to and even beyond the recommended dose in what may be a futile effort to achieve a response (futile in the sense that the patient may have an illness that is unresponsive to the mechanism of action of the drug being prescribed). Even if the patient does not respond to the drug, there is a chance that he or she may remain on it with more drugs being added to try to induce a response. In addition, one out of every two responders in usual clinical registration trials of antidepressants is responding to nonspecific treatment factors, subsumed by the phrase "placebo response."¹⁷ Consequently, there is also no way to know whether the patient in clinical practice who has responded has done so specifically because of the drug or because of nonspecific treatment factors (i.e., "placebo" response).^18,19

The history of syndromic diagnoses in medicine indicates that they tend to be subdivided into more pathophysiologicially and/or pathoetiologically defined diseases as more knowledge is gained. This process has already occurred with certain psychiatric syndromes and is likely to continue. For example, conditions such as pellagra, general paresis of the insane, and a variety of druginduced conditions were once lumped together with other psychiatric conditions. The fact that these conditions can present as other psychiatric disorders but, in fact, require different treatments because they have different etiologies is the reason for the specific exclusion criteria included in DSM-IV diagnostic criteria. Such subdivisions of psychiatric syndromes into smaller and more refined disorders are likely to accelerate as a result of the human genome project and advances in neuroscience research, but that is not the case now. That could limit the size of the market for future psychiatric drugs and thus affect the economics of drug development.

Inclusion and Exclusion Criteria

The goal of clinical registration trials is to produce sufficient data to permit the submission of a successful NDA to a drug regulatory agency such as the FDA. Since drug development is a business, that goal must be achieved in a manner consistent with achieving the highest possible return on investment.^20-23 That, in turn, means limiting the cost of development while maximizing the market.

As noted above, this column is part of a series on the drug development process.² Since other columns in that series have provided an overview of how the drug development process and subsequent marketing efforts are structured to limit cost while maximizing return, this column will focus on just one aspect of limiting cost, the screening process that determines which patients can go into a clinical trial and what impact that has on the generalizability of the results to everyday clinical practice. For brevity, this discussion will focus on clinical trials of antidepressants but the general theme is applicable to clinical trials for other types of disorders such as anxiety disorders, psychotic illnesses, and dementias.

Table 1 lists the typical inclusion and exclusion criteria for an antidepressant clinical trial. The criteria are generally applicable to all clinical trials in psychiatric disorders, with inclusion criterion #1 and exclusion criteria #1 and #2 being modified for the specific conditions being treated (e.g., schizophrenia).

Readers, if they are practicing clinicians, can reflect over the last 100 patients that they have treated with an antidepressant and consider how many would meet these criteria and thus be eligible for such a trial. When this question has been posed at lectures with hundreds of such clinicians in attendance, the answer has been an exceedingly small percentage (e.g., 5% or less). That is particularly true for psychiatrists because they often treat patients who have not responded to earlier treatment by a primary care provider. Yet most of the systematic data about the safety, tolerability, and efficacy of a drug is limited to patients who meet the criteria listed in Table 1.

There are at least two reasons why these criteria are used. The first is to enhance the likelihood of seeing a drug-placebo difference, the current FDA requirement for the successful registration of a new antidepressant. The second is to reduce the likelihood of adverse events (AEs), particularly severe or serious AEs, during the clinical trial. The FDA definition of an adverse event is "any unfavorable and unintended sign (including an abnormal laboratory finding), symptom or disease temporally associated with the use of a medicinal (investigational) product, whether or not considered related to the medicinal (investigational) product."²⁴ The FDA definition of a serious AE is given in Table 2.

AEs are part of everyday life, and they also sometimes occur because of the illness being treated. For example, a certain percentage of depressed patients will commit suicide (even though they deny being suicidal) or become psychotic or manic or have a serious accident due to inattention caused by their depressive condition. If the patient has a comorbid medical illness or is taking a concomitant medication, he or she may have an AE secondary to that fact. The likelihood of such an AE occurring is increased if the patient’s medical condition is unstable or has significant potential for being unstable. This is particularly problematic if the AE may be serious (e.g., a seizure in an epileptic patient). The risk of an AE is also increased if a concomitant medication may adversely interact with the study drug or if the patient has a significant history of drug intolerance or allergies. All of these factors compromise the ability to determine whether the AE is due to the investigational drug or to one or more of these other factors.

In addition to dividing AEs by severity, the FDA also divides AEs into three categories on the basis of frequency:

  frequent:   incidence of 1/100 or greater
  infrequent: incidence between 1/100 and 1/1000
  rare:       incidence less than 1/1000.

The best way to determine whether the risk of an AE is due to the drug is to statistically compare the incidence in the group treated with the drug with the incidence in the group treated with placebo (i.e., the baseline rate). Most successful registration trial packages for recently approved antidepressants include placebo-controlled, head-to-head comparison of several hundred patients on placebo versus several hundred patients on the new drug; the total number of patients on the drug in all trials is generally a few thousand. Such numbers generally provide only enough power to test for the difference in frequent AEs.

A complicated problem has to do with serious and rare AEs, such as agranulocytosis, liver failure, seizures, mania, or suicide. If an event is rare on both the investigational drug and placebo, then there is no power to determine whether the risk on the drug is truly greater than would be expected by chance (i.e., placebo). For example, 3 patients out of 3,500 on an investigational drug (all exposures) may have developed liver failure compared with 1 out of 1,000 on placebo. The rate on drug is numerically three times higher, but is not statistically different from the rate on placebo because of the rare nature of the event.

These considerations are an important reason for the exclusion criteria listed in Table 1. Such exclusion criteria are in part intended to reduce the occurrence of serious AEs during the registration trials with a new agent, because the occurrence of such AEs can significantly complicate the approval process. Moreover, the occurrence of serious AEs, even if of uncertain relationship to the new drug, may be reflected in the package insert, which can limit market acceptance of the drug. A company may even decide not to market a drug if FDA approval is contingent on the package insert containing a "black box warning," as was the case with the approvable but not marketed antipsychotic, sertindole.

Special Population Trials

As the drug progresses through its development and is thus more likely to be marketed, an attempt is made to address some of the variables specifically omitted by the criteria in Table 1. These later studies are often a variation of the early phase I studies, but, instead of being done in normal volunteers, they are performed in special at-risk populations, such as subjects with varying degrees of renal or hepatic impairment. These studies are often small but focused on the same outcome variables as the early phase I studies: safety, tolerability, and pharmacokinetics rather than efficacy. Small studies, with variable designs such as phase I or open or double-blind efficacy trials, may also be done in elderly individuals; however, such studies are generally not powered to prove a difference in efficacy between drug and placebo.

Traditionally, drug-drug interaction studies have also been done late in a drug’s development; however, such studies are now being performed earlier in the development process as a result of the improved ability to predict and test for drug-drug interactions, particularly those mediated by cytochrome P450 enzymes.^25-33

Nevertheless, there are several limitations to these studies. First, only a small number of these studies (perhaps six) are done as part of the development package. Second, these studies are almost always limited to two drugs, the investigational agent and a potential perpetrator or victim drug. In contrast to this simple co-pharmacy situation, one third of patients who are taking an antidepressant in either a primary care or an outpatient psychiatric setting are receiving three or more medications in addition to their antidepressant.³⁴ A recent survey found that 10% of all American over the age of 18 were on five or more prescription drugs in the week preceding the survey.³⁵ Thus, clinically significant drugdrug interactions may be discovered only after a drug has been on the market for a period of time.³⁶

Relevance to Clinical Practice

Obviously, there is an enormous difference between the usual patient in clinical practice on an antidepressant and the usual patient in the registration trials that led to the drug’s approval. The usual patient on an antidepressant in clinical practice frequently has one or more comorbid conditions, including other psychiatric illnesses, substance abuse or dependence disorders, or general medical conditions. Such a patient is therefore frequently on more than one other medication. Depending on the nature of the practice, the majority of a practitioner’s patients may be either younger or older than the usual patient in the registration trials. Relative to the last point, the number of patients over 60 years of age is generally small in most registration trials of antidepressants, even if a upper age cutoff is not specified in the inclusion or exclusion criteria.

All of the factors discussed in the preceding paragraph are subsumed under variable 3 in equation 1: age, disease, and environment (e.g., concomitant medications). Each factor can individually cause the patient’s response to vary from the usual patient in the registration trial. The more the practitioner’s patient differs from the usual patient in the registration trials, the greater the likelihood that there will be a clinically significant difference in the expected response. This is one reason for the adage of "start low and go slow" when treating the elderly with any drug, but particularly a new drug.

In the final analysis, the astute and careful clinician is the critical safeguard when a new drug is introduced to the market. The practitioner, whether he or she realizes it, is frequently using a new drug in novel circumstances that were never tested during the registration trials. The goal of this column was to summarize some of the reasons for this statement in the hope that it might provide a useful perspective for readers who are incorporating a new drug into their practice or treating a patient who is receiving a new drug prescribed by another practitioner.

References

1. 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 9(4), July 2003
2. Preskorn SH, Modern drug development and the Human Genome Project. (collection of 4 columns published in the Journal of Psychiatric Practice). Accessed June 25, 2001
3. Preskorn SH, The stages of drug development and the human genome project: Drug discovery. Journal of Psychiatric Practice 2000;6:341-4
4. Offord SJ, Wong DF, Nyberg S., The role of positron emission tomography in the drug development of M100907, a putative antipsychotic with a novel mechanism of action. J Clin Pharmacol 1999;Aug (suppl):17S-24S
5. Kuhlmann J, Muck W., Clinical-pharmacological strategies to assess drug interaction potential during drug development. Drug Saf 2001;24:715-25
6. Preskorn SH, Section 3. The rationale for and clinical implications of dose-response curves. (five columns published in the Journal of Practical Psychiatry and Behavioral Health in 1996)
7. Preskorn SH, Reproducibility of the in vivo effect of the selective serotonin reuptake inhibitors on the in vivo function of cytochrome P450 2D6: An update (part II) Journal of Psychiatric Practice 2003;9:228-36
8. Evans WE, Johnson JA., Pharmacogenomics: The inherited basis for interindividual differences in drug response. Annu Rev Genomics Hum Genet 2001;2:9-39
9. Herrmann SM, Nicaud V, Tiret L, et al., Polymorphisms of the beta2-adrenoceptor (ADRB2) gene and essential hypertension: The ECTIM and PEGASE studies. J Hypertens 2002;20: 229-35
10. Gingrich JA., Mutational analysis of the serotonergic system: Recent findings using knockout mice. Curr Drug Target CNS Neurol Disord 2002;1:449-65
11. Siest G, Ferrari L, Accaoui MJ, et al., Pharmacogenomics of drugs affecting the cardiovascular system. Clin Chem Lab Med. 2003;41:590-9
12. Brunner HG, Nelen M, Breakefield XO, et al., Abnormal behavior associated with a point mutation in the structural gene for monoamine oxidase A. Science 1993;262:578-80
13. Schuback DE, Mulligan EL, Sims KB, et al., Screen for MAOI mutations in target human groups. Am J Med Genet 1999;88: 25-8
14. Shih JC, Chen K, Ridd MJ., Monoamine oxidase: From genes to behavior. Annu Rev Neurosci 1999;22:197-217
15. Shih JC, Thompson RF., Monoamine oxidase in neuropsychiatry and behavior. Am J Hum Genet 1999;65:593-8
16. 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 8(3):170-177, May 2002
17. Janicak P, Davis JM, Preskorn SH, et al., Treatment with antidepressants. In: Principles and practice of psychopharmacotherapy, 3rd edition. Philadelphia: Lippincott Williams & Wilkins; 2001: 215-325
18. Preskorn SH, A dangerous idea. J Pract Psychiatry Behav Health 1996;2:231-4
19. Preskorn SH, Patients don’t fail treatments, treatments fail patients. J Pract Psychiatry Behav Health 1997;3:165-8
20. Kuhlmann J, Alternative strategies in drug development: Clinical pharmacological aspects. Int J Clin Pharmacol Ther 1999;37:575-83
21. Chaikin P, Rhodes GR, Bruno R, et al., Pharmacokinetics/pharmacodynamics in drug development: An industrial perspective. J Clin Pharmacol 2000;40:1428-38
22. Caldwell GW, Ritchie DM, Masucci JA, et al., The new pre-preclinical paradigm: Compound optimization in early and late phase drug discovery. Curr Top Med Chem 2001;1:353-66
23. DiMasi JA., The value of improving the productivity of the drug development process: Faster times and better decisions. Pharmacoeconomics. 2002;20 (Suppl 3):1--10
24. ICH Guideline for Good Clinical Practice Title 21: Food and Drugs. Code of Federal Regulations, April 1, 2003
25. Rodrigues AD, Rushmore TH., Cytochrome P450 pharmacogenetics in drug development: In vitro studies and clinical consequences. Curr Drug Metab 2002;3:289-309
26. Bu HZ, Magis L, Knuth K, et al., High-throughput cytochrome P450 (CYP) inhibition screening via a cassette probe-dosing strategy. VI. Simultaneous evaluation of inhibition potential of drugs on human hepatic isozymes CYP2A6, 3A4, 2C9, 2D6 and 2E1. Rapid Commun Mass Spectrom 2001;15:741-8
27. Masimirembwa CM, Thompson R, Andersson TB., In vitro high throughput screening of compounds for favorable metabolic properties in drug discovery. Comb Chem High Throughput Screen 2001;4:245-63
28. Riley RJ, Parker AJ, Trigg S, et al., Development of a generalized, quantitative physicochemical model of CYP3A4 inhibition for use in early drug discovery. Pharm Res 2001;18:652-5
29. Bonnabry P, Sievering J, Leemann T, et al., Quantitative drug interactions prediction system (Q-DIPS): A dynamic computerbased method to assist in the choice of clinically relevant in vivo studies. Clin Pharmacokinet 2001;40:631-40
30. Kariv I, Fereshteh MP, Oldenburg KR., Development of a miniaturized 384-well high throughput screen for the detection of substrates of cytochrome P450 2D6 and 3A4 metabolism. J Biomol Screen 2001;6:91-9
31. Gao F, Johnson DL, Ekins S, et al., Optimizing higher throughput methods to assess drug-drug interactions for CYP1A2, CYP2C9, CYP2C19, CYP2D6, rCYP2D6, and CYP3A4 in vitro using a single point IC(50). J Biomol Screen 2002;7:373-82
32. Bjornsson TD, Callaghan JT, Einolf HJ, et al., and Pharmaceutical Research and Manufacturers of America Drug Metabolism/Clinical Pharmacology Technical Working Groups, The conduct of in vitro and in vivo drug-drug interaction studies: A PhRMA perspective. J Clin Pharmacol 2003;43:443--69
33. Kremers P., Can drug-drug interactions be predicted from in vitro studies?. Scientific World Journal 2002;2:751-66
34. Preskorn SH, Use of antidepressants with other medications. In: Preskom SH. Outpatient management of depression. Caddo, OK Professional Communications; 1999:179-91
35. Kaufman DW, Kelly JP, Rosenberg L, et al., Recent patterns of medication use in the ambulatory adult population of the United States: The Slone Survey. JAMA 2002;287:337-44
36. Redfern WS, Wakefield ID, Prior H, et al., Safety pharmacology -- A progressive approach. Fundam Clin Pharmacol 2002;16: 161-73