Clinical Trials: Out of the Laboratory and Into the Bathroom Cabinet: Designing and Implementing Clinical Trials
Once a drug candidate has been identified and undergone study in the laboratory and animals, it enters clinical trial, usually composed of 3 phases. Generally, as the drug candidate goes from phases 1 to 3, the number of volunteers exposed to it increases and the duration of study lengthens. Each phase serves a unique purpose, and each builds on the knowledge gleaned from the earlier research.
Not all clinical trials involve an investigational drug. Some involve medical devices (for example, an insulin pump) and some examine clinical management strategies. The recently launched SMART study (Strategies for the Management of Anti-Retroviral Therapy), which pits early anti-HIV treatment against delayed anti-HIV treatment, is an example of such a study. However, this article focuses on the clinical trial of medications for treating HIV infection.
Although often given long titles and identified by imposing acronyms, a clinical trial of a drug candidate is simply a scientific way to answer specific questions about the new therapy. The questions usually differ from one phase of research to the next. For potential study volunteers, it's important to remember that the question a study asks may not be well thought out or the answer unimportant. All clinical studies are not created equal. Also, most clinical trials of anti-HIV drugs are intended only to obtain regulatory approval for marketing the drug. The trials do not necessarily tell patients and clinicians how best to use a drug. For example, abacavir sulfate (Ziagen) received regulatory approval when research showed that it lowered viral load. But many clinicians are still unsure of whether to use abacavir as part of a first treatment regimen or to save it for later, or whether to combine it with a protease inhibitor or a non-nucleoside reverse transcriptase inhibitor.
In a phase 1 study, researchers usually expose only a handful of volunteers -- sometimes not more than 10, and rarely more than 80 -- to the new therapy. The reason for this is simple: a phase 1 study marks the first time humans receive an investigational drug and researchers want to minimize harm if the therapy proves unsafe. A phase 1 study allows investigators to determine the side effects and safe dose of a new treatment. Put as a question, the phase 1 study asks: "Is this drug safe for people to take, and if so, at what dose?" Phase 1 studies almost never ask, "Is this drug effective?" It can be tempting for research participants, especially if they are eagerly awaiting a new treatment, to just assume that any new drug is effective. But the question of effectiveness -- and the answer to it -- comes later. Although essential to the process of clinical research, a phase 1 study carries the greatest risk to study volunteers.
If the results from the phase 1 study indicate that a drug is safe, it advances to phase 2. In phase 2, investigators may expose a couple of hundred people to the experimental therapy, allowing for further characterization of the drug's safety profile. Phase 2 usually marks the first time that researchers examine whether a drug is efficacious. In the context of HIV clinical therapy research, efficacy means the ability of a drug to produce a therapeutic change at the maximum dose tolerated by humans.
Specifically, investigators are usually interested in changes in the CD4 T cell count or viral load or both. Changes in these surrogate markers are usually the study endpoints, and endpoints are the measurements by which a drug's efficacy is judged. Length of survival and a decrease in disease and disability are the endpoints that truly interest most physicians and patients. But measuring those kinds of clinical endpoints may take years. (The SMART study, for example, will measure clinical endpoints and will last from 6 to 8 years.) So to speed the process along, researchers often use surrogate markers instead of clinical indices as study endpoints. A beneficial change in surrogate markers is thought to predict or stand in for a beneficial change in clinical outcome. Conversely, a negative change in surrogate markers -- a decline in CD4 T cell count or a rise in viral load -- is thought to predict a negative change in clinical outcome.
That's usually true, but not always. Sometimes the therapy itself may alter the predictive value of a surrogate marker. For example, Steven Deeks, MD, has shown that HIV-infected patients who remain on a protease inhibitor do better clinically than their viral loads would indicate. Still, surrogate markers are generally accurate and their use dramatically reduces the amount of time needed to study potential new anti-HIV treatments.
If all goes well, an investigational drug will end up in a phase 3 study where several hundred people receive it for 6 months or longer. The purpose of this phase is to determine the efficacy of an agent in a large number of people. Phase 3 studies also allow for the detection of side effects that may not have emerged in the earlier, small studies. All phase 3 studies involve a control group. The control group does not receive the investigational agent but instead gets a placebo or, more likely nowadays, a standard of care regimen. In HIV care, standard of care treatment might include a protease inhibitor and 2 nucleoside analogs. The control group serves as a benchmark against which the investigational therapy is judged.
For phase 3 studies, investigators usually recruit patients who share certain characteristics in common. For instance, investigators may seek patients who have never before taken anti-HIV drugs, or who have a CD4 count above, or a viral load below, a specific cut-off value. The study volunteers are then randomized -- or assigned by chance -- to either the study group or the control group. Theoretically, randomization eliminates scientific bias and ensures that the only difference between the groups is the treatment they receive. In this way, researchers can compare the effects of the new treatment against the standard of care in a study group whose characteristics are similar to those of the control group. The requirements for participation in a study are known as the inclusion/exclusion criteria. These requirements are spelled out in the study protocol, a set of written instructions for the conduct of the study.
Once a study is complete, researchers analyze the results to see if the investigational therapy was inferior, equivalent, or superior to the standard of care. The difference in results, if any, is determined by comparing the statistical significance of changes in the study endpoints between the study group and the control group. For example, if the study endpoint was a change in viral load, investigators may ask, "What was the average change in the study group's viral load compared to the control group's? Is this difference statistically significant?"
Since the volunteers in both groups shared similar characteristics at the start of the study, their baseline viral loads should have been about equal. And since they were assigned by chance to either the study group or control group, any significant changes at the end of the study in their viral loads should be attributable to the treatment they received -- the treatment being the only difference between them. If, at the end of the study, the viral loads of the 2 groups are roughly the same, the researchers may conclude that the new therapy is equivalent to existing treatments: no better, no worse.
If the viral load of the study group is significantly lower, statistically speaking, than that of the control group, the researchers may conclude that the new treatment is superior to existing treatments. There is a potential catch, however. All statistically significant changes are not necessarily clinically significant. In other words, a statistically significant difference may not actually matter for the health of patients. To see why, consider this analogy. Suppose researchers offer 2 groups of 100 patients a drug designed to increase their earning power. Suppose that at the end of the trial, the study group has earned 5 cents and the control group has earned only 1 cent. Is this difference statistically significant? Yes, it is. But is it economically significant? No, not really. In the US economy, whether people have a nickel or a penny matters very little. Either way, they aren't buying much. Similarly, changes between groups in viral load could be statistically significant without being clinically significant.
Also, changes in viral load or CD4 T cell count do not tell the whole story. Other important considerations are a drug's convenience, tolerability, and safety. Say an investigational drug dramatically reduced viral load but had to be taken 5 times a day on an empty stomach, caused severe nausea, and put the patient at increased risk for liver cancer? Is it a good drug?
Often investigators examine the difference between a drug's treatment effect (its ability to reduce viral load or raise T cell count) and its clinical effect (which includes convenience and tolerability) by performing 2 sets of analysis. In one type of analysis known as on treatment, researchers consider only the treatment effect of a drug and count only those patients who actually take the medication. In other words, if a patient stays on treatment, regardless of side effects or inconvenience, how well does the medication affect the surrogate markers? In another type of analysis, known as intent to treat, researchers include the consequences of a drug's side effects and dosing schedule and count all the patients in a study, whether they take the medication or not. The difference in results between these 2 types of analysis can be sharp.
Say 500 hundred patients are given experimental drug A. If 250 of them take the drug faithfully and it lowers their viral loads to undetectable, an on-treatment analysis will show 100 percent efficacy, even though the other half of the patients did not take the drug because it made them sick. On the other hand, an intent-to-treat analysis -- designed to capture the "real world" results of a medication -- will count all the patients, including those who couldn't stomach the drug's side effects. By this method, the experimental drug will show efficacy of 50 percent, having been credited for working only in those patients who could take it.
For patients deciding whether to participate in a clinical trial, it is important to know what phase of study the drug is in and whether those who are assigned to the control arm will receive a placebo or standard of care. Also, will the study be blinded or open label? In other words, will the volunteers know what arm of the study they are in? In a blinded trial, the volunteers are not told whether they are in the study arm or the control arm. In an open-label trial, they know. Blinded trials adjust for bias on the part of study volunteers -- for example, when volunteers report side effects caused by the medication even though they're actually receiving a sugar pill -- but blinded trials also increase the uncertainty of participating in a study.
Another key question is, "What exactly do the researchers hope to learn from this trial?" At the heart of every clinical trial, regardless of the phase of study, there is a scientific question. Informed study volunteers find out what the question is and decide for themselves whether they think it is worth answering. The informed consent document explains the purpose of the study as well as the risks and benefits the study offers to volunteers.
In reviewing the results of a clinical trial, it is important to know what population of patients the study included. The results of a study can only be generalized to patients who share the characteristics of the study volunteers. For example, if a study included only volunteers who were naive to anti-HIV drugs, the results cannot be generalized to patients who are heavily experienced with anti-HIV drugs. Similarly, a drug that produces good results in volunteers who started out with 500 CD4 T cells may not work as well in patients whose T cell counts are less than 100. The characteristics of the patients who participated in the study are determined by reviewing the inclusion/exclusion criteria.
Finally, an essential issue is whether the results of a study are not only statistically significant, but also clinically significant. What endpoints did the researchers use for evaluating a drug's efficacy? Are results reported from an on-treatment analysis or an intent-to-treat analysis? Clinical trials are most helpful when they ask important questions, are well designed, recruit informed volunteers, and report clinically meaningful information.
GlossaryPhases 1 to 3
: Clinical studies are usually divided into 3 phases, with each phase serving to answer a specific question about an investigational therapy.
Study volunteers: In the context of HIV clinical research, study volunteers are men and women with HIV infection who meet a study's inclusion/exclusion criteria and agree to participate in the trial.
Surrogate markers: Laboratory tests, usually CD4 T cell count and viral load, that are used to estimate the health benefit of an investigational drug.
Study endpoints: The measures used to determine whether an investigational drug works. In HIV clinical research, the study endpoints are usually a change in surrogate markers.
Control group: A group of patients in a study who receive a "sugar pill" or conventional treatment rather than the experimental therapy. A control group is necessary to determine whether an investigational drug offers any benefit over existing therapies.
Placebo: The formal name for a "sugar pill." An inactive substance.
Standard of care: The level or type of care generally given to people with a particular condition.
Study group: A group of patients in a study who receive the experimental drug.
Inclusion/exclusion criteria: The list of characteristics that define who may and may not participate in a clinical study.
Protocol: A written description of a study and a guide to its conduct. Protocols are usually lengthy documents, sometimes running hundreds of pages in length.
Baseline: Refers to anything present at the start of a study. For example, if a patient enters a study with a viral load of 50,000 copies, this is the baseline viral load.
Treatment effect: The effect of an investigational drug on study endpoints (e.g., viral load or CD4 T cell count).
Clinical effect: The effect of an investigational drug on factors other than the study endpoints, including, for example, quality of life.
On treatment: A type of analysis in which researchers count only those patients who actually took the investigational drug for the study duration. This type of analysis excludes patients who failed to take the medicine, whether or not they had a good reason for doing so. On-treatment analysis is best for determining the treatment effect of a new therapy.
Intent to treat: A type of analysis in which researchers count every patient who participated in the study of an investigational drug. This type of analysis includes patients who failed to take the medicine, whether or not they had a good reason for doing so. Intent-to-treat analysis is best for determining the clinical effect of a new therapy.
Blinded: A type of study in which the volunteers are not told whether they are receiving the investigational drug or standard of care. If even the doctor does not know which patients are receiving the investigational drug, the study is called "double blinded." Blinded studies help to reduce bias.
Open label: A type of study in which the volunteers are told whether they are receiving the investigational drug or standard of care.
Informed consent: A document that explains, in easy-to-understand language, the purpose of a study and the risks and benefits it offers for volunteers. Every volunteer in a study should receive a copy of the informed consent document.
Generalized: The act of having taken results from a study and applied them to people who did not actually participate in the study. Results from a clinical trial may be generalized only to patients who share the characteristics of the study volunteers (e.g., similar viral loads or treatment history).
How Do They Do That?
A Look at Structure-Based Drug Design
Where do potential anti-HIV drugs come from in the first place? Before any drug candidate reaches clinical trial, researchers must first identify and test it in laboratory studies. How do scientists find these candidates, and what's the process for testing them?
HIV has 3 enzymes that perform important work for the virus. (An enzyme is a protein that catalyzes, or facilitates, chemical reactions.) These enzymes are targets for therapy, and to date researchers have identified and successfully tested inhibitors of two of HIV's enzymes, namely reverse transcriptase and protease. But an inhibitor of the third enzyme, integrase, has so far eluded researchers. Several research groups are still looking for an integrase inhibitor, including one at the University of Houston, led by Kurt Krause, M.D., Ph.D.
To find an inhibitor for a viral enzyme, researchers start by studying the structure of the enzyme itself. Using sophisticated technology, samples of the enzyme are examined to determine its shape and to identify the point where an inhibitor might attach. This point is known as the "active site." Once the structure of the enzyme is identified, that information is fed into a computer which sorts through a database of hundreds of thousands of small molecules, looking for one that fits the enzyme. Several of these "virtual" molecules might fit, some better than others. The computer scores a molecule's affinity for the enzyme, ranking it as high, medium, or low. A high-affinity molecule may become a "lead compound," at which point chemists begin to make it. Other scientists will then test the lead compound in the laboratory against actual viral enzymes. (See figure below.)
If it works, the molecule will undergo further testing to estimate whether it is safe for humans to receive, and if so, whether it is bioavailable. (Bioavailability refers to how well a drug is absorbed by the human body.)
According to Krause, the problem with finding an inhibitor for HIV integrase is two-fold. First, the structure of the integrase protein complex isn't fully understood, and the "active site" where a drug might attach is shallow. Reverse transcriptase, in contrast, has a shape similar to a hand with the fingers drawn in, and the palm -- the active site -- serves as a deep pocket to which drug can stick.
In the coming years, Krause (who is also interested in finding treatments for other infectious diseases, including tuberculosis) estimates that advances in technology and improvements in efficiency will reduce the cost of drug discovery by several million dollars. This should help to reduce the cost of treatments and make them more widely available, especially in the developing world.
This article was provided by The Center for AIDS. It is a part of the publication Research Initiative/Treatment Action!. Visit CFA's website to find out more about their activities and publications.