THE MAJOR TASKS OF EACH PHASE OF CLINICAL DEVELOPMENT

Traditionally clinical development is divided into three phases, though these designations have no clearly-defined meaning. Mechanically pushing a compound through each phase may distract from the most important task of clinical drug development: individualizing the design and sequencing of studies to optimally advance the compound.

Roughly speaking, Phase I refers to small (8-60 subjects), usually short studies designed to elucidate the drug's basic safety profile, pharmacokinetics, and sometimes pharmacodynamics. Phase I usually involves healthy volunteers. The drug is typically first administered to patients in Phase II trials, which are of medium size (50 250 patients). Such studies may provide the first evidence of efficacy, identify the main side-effects in patients, and determine the clinically relevant dose range. Phase III studies are generally much larger (300 -- 1000+ patients). They are designed to refine dosing and provide evidence of efficacy and safety in a more diverse group of patients than Phase II, mimicking actual clinical practice as much as possible.

EXPLANATION OF POWER AND P VALUES

The chance that a trial outcome is really a false positive result is called a P value, and a trial must show that a treatment has a positive effect with a P ≤ 0.05 to be considered statistically significant (i.e. 5% or lower chance that the result is due to chance). Clinical trials are often described, for example, as being 90% powered to show a 20% treatment benefit versus placebo. Power is defined as the probability that if a drug can yield a meaningful difference in a clinical endpoint, the trial will show it with P ≤ 0.05. Important variables for calculating power are the number of subjects in the trial and the definition of meaningful difference. The less dramatic a drug's effect, the larger the trial must be to achieve 90% power.

PHASE I

Phase I lays the groundwork for the entire subsequent development. Skimping on vital Phase I studies in a rush to proceed to Phase II may cripple the clinical program and ultimately undermine the NDA. Too often during the design of critical Phase III studies one hears someone say, "if only we knew this about the compound," referring to such things as the highest well-tolerated dose, the kinetics of dosing three versus two times daily, or another parameter that could have been easily obtained in a Phase I study.

The main task of Phase I is to determine whether the drug merits further clinical testing and, if so, to provide key information necessary for designing these trials. Establishing the drug's safety profile is critically important; unacceptable toxicity will prevent even the most efficacious drug from being approved. Of course what constitutes unacceptable toxicity may be very different for an antihistamine and an anti-cancer compound. Escalating single doses, then multiple doses, should be administered until either the unit dose is ridiculously high or, more commonly, dose-limiting side-effects occur. It is vital at an early stage to know the most common side-effects associated with a new drug and how these are related to dose.

The second major Phase I task is to determine the drug's basic pharmacokinetic (pK) profile. Carefully designed studies should establish whether the formulation chosen for clinical development can produce therapeutically-relevant plasma drug concentrations. For example, unless reasonable blood levels can be achieved when the drug is given orally at the highest safe dose, it does not make sense to use such a formulation in a proof of concept study. Either the company must develop a new oral formulation or select an alternative route of administration that results in better bioavailability. Early pK studies also determine the drug's half-life, which will help establish how frequently the drug should be dosed. Occasionally, insurmountable pK issues identified in Phase I result in the termination of a project.

A third Phase I task, applicable to some but not all drugs, is to characterize the compound's basic pharmacodynamics (pD), usually as a function of plasma drug level. A drug designed to lower blood sugar, depending on its mechanism of action, might be studied in either normal volunteers or people with diabetes. The goal would be to establish the relationship between drug level and plasma glucose. Important information may be obtained even from a small Phase I study on the lowest dose of drug that is associated with the desired benefit; in such a case, both the lowest and highest doses for subsequent efficacy testing could be elucidated in a single study. In the case of a cytotoxic drug designed for the treatment of malignancy, study subjects would necessarily be people with cancer. Pharmacodynamics would be measured using a surrogate for antineoplastic effect, possibly levels of certain lymphocyte populations.

It is important to recall that Phase I refers to a type of study, not a chronological order. Many of the Phase I studies that regulatory authorities require for approval are best performed only after it is certain that an NDA will be filed, including studies on drug interactions, fed/fasted pharmacokinetic differences, pK and pD as a function of age and gender, and kinetics in renal and hepatic failure. There is no point in doing such trials before proof of concept is established since the results are seldom relevant to the patient population enrolled in early efficacy studies.

PHASE II: EMPHASIZING PROOF OF CONCEPT

The most important traditional task of Phase II is to establish proof of concept (POC), that is, the first credible evidence in the target population that the drug actually does what it is being developed to do. For certain conditions (e.g. migraine), POC can sometimes be obtained in a Phase I study, whereas for others (e.g. anxiety) it is not achieved until Phase III. Cancer is an example of a field where Phase III results often fail to live up to expectations set by positive Phase II data, even when the Phase II endpoint is survival. Phase II cancer studies sometimes lack proper placebo control arms, are not blinded, or are not randomized properly. Instead, these elements of a proper trial design (discussed below) are reserved for Phase III studies while the Phase II studies use historical or case-matched controls instead of placebo arms. Human bias in trials is very real and significant -- without proper placebo controls, randomization, and blinding, trial results cannot serve as Proof of Concept.

The validity of the POC trial depends on how well it is designed, conducted, and analyzed. A properly done study that is clearly negative is a strong argument for terminating the development of a drug, whereas a positive outcome will lead to a huge investment of resources. Because the stakes are so high, every effort must be made to ensure that the POC results are trustworthy. Perhaps the most common, and ultimately the most costly, mistake emerging companies make is to under-fund and underpower a POC trial "because we don't yet know if the drug works." A so-called exploratory study, the clinical research equivalent of a toe in the water, seldom provides useful information.

Below are some key elements of a robust POC trial.

• It must be designed to answer an appropriate clinical question. This is not the time to do mechanistic studies, no matter how fascinating the information may be. The problem is that mechanistic endpoints (e.g. a measure of how the drug alters physiology) do not reliably predict actual clinical utility. For example, in a POC study for the treatment of irritable bowel syndrome, it would be inappropriate for the primary endpoint to be some aspect of gut motility. Rather, it must be a validated measure of clinical response. If a surrogate marker is employed, it should be a well-documented predictor of meaningful clinical effect (e.g. tumor regression in cancer trials).
• It should be placebo controlled, randomized, and double-blinded.
• It should involve a patient population that resembles that which will be the ultimate market for the drug.
• It must explore a wide enough range of doses to ensure that a negative result is not due to under-dosing.
• It should be designed to begin the process of identifying the optimal clinical dose(s), including the documentation of dose-response relationships.
• It must be sufficiently powered so there is little likelihood of erroneously concluding that the drug does not work. Typically the power of a POC study is set at 90%, (i.e. there will be only a 1 in 10 chance of a false negative result).
• It should characterize the nature and frequency of the most prominent side-effects to be expected in actual patients.

Phase II trials also help quantify a drug's benefit versus placebo so that enough patients are recruited in Phase III trials to have a good chance (90%) of showing that the observed benefit is statistically significant. If a drug shows a weak benefit to the patient in Phase II trials, many patients will be required in Phase III to demonstrate that this weak benefit is, in fact, statistically significant (P ≤ 0.05).

PHASE III: EMPHASIZING GENERAL PRINCIPLES OF CLINICAL STUDY DESIGN

The major task of Phase III is to conduct two independent clinical trials that conclusively prove that the compound is effective and safe. Such studies, required for the regulatory approval of most drugs, are called pivotal trials. Their design and execution are critical since the success of the NDA depends to a great extent on their outcome. The specifics of Phase III study design are highly dependent on the nature of the molecule being developed, the therapeutic target, the particular efficacy endpoints employed, and the results of earlier clinical studies.

It is generally recognized that meaningful clinical trials must be appropriately controlled, randomized, and blinded.

Despite the Helsinki Declaration on Human Experimentation, the control group in most clinical trials receives a placebo rather than, as required, "the best proven therapeutic method." This is because a placebo control generally makes it much easier to show that the experimental drug is effective. To maximize the likelihood that patients receiving the experimental drug and those receiving the control drug are not meaningfully different in any other way, treatment must be randomly assigned, usually by a standard computer-generated paradigm. Finally, to prevent knowledge of the patient's treatment from influencing patients and investigators, both must be blinded to treatment assignment, i.e., the experimental and control medications must appear identical in all respects. The trial is unblinded for analysis only after the study is completed.

The following are under-appreciated key principles:

• Begin with a clear, simple objective: Because clinical trials are so costly and time-consuming, there is a strong temptation to try to answer many questions in a single study. For example, scientists may be tempted discern a drug's mechanism of action by measuring various physiologic parameters during the trial. When employed judiciously such ancillary measures are acceptable. If they dominate the trial, however, they can be distracting.
• Design and power the study around a single primary question: The most important question that the study seeks to answer should be operationalized in the primary endpoint. The success of the study depends on whether the primary endpoint is both significantly different, both clinically and statistically, between the experimental treatment and the control.
• A clinical trial must be designed as one in a chain of studies: The study should both take into account the results of previous trials and produce data suitable for refining the features of subsequent ones (e.g. selecting patient population, trial length, drug doses).
• Answer the essential questions without generating unnecessary data. All data generated during clinical development will be included in the NDA submission and may find its way into the package insert. You may not like the answers you get to questions you did not have to ask.